Diagnostics

12 pages, 913 KB

Open AccessCase Report

Atypical Pruriginous Pustular Eruption Preceding Locally Advanced Rectal Cancer: A Case Report and Gut–Skin–Tumour Axis Hypothesis

by Monica Manciulea (Profir), Luciana Alexandra Pavelescu and Sanda Maria Crețoiu

Diagnostics 2026, 16(11), 1592; https://doi.org/10.3390/diagnostics16111592 (registering DOI) - 22 May 2026

Abstract

Background and Clinical Significance: Cutaneous paraneoplastic phenomena are infrequently characterised in colorectal cancer (CRC), and chronic pruriginous inflammatory eruptions in particular have received limited attention. In older adults, persistent treatment-resistant dermatoses of unclear aetiology may represent overlooked extraintestinal diagnostic clues to occult malignancy, [...] Read more.

Background and Clinical Significance: Cutaneous paraneoplastic phenomena are infrequently characterised in colorectal cancer (CRC), and chronic pruriginous inflammatory eruptions in particular have received limited attention. In older adults, persistent treatment-resistant dermatoses of unclear aetiology may represent overlooked extraintestinal diagnostic clues to occult malignancy, including potentially curable CRC. Faecal immunochemical testing (FIT) for occult bleeding is a low-cost, non-invasive tool whose role outside conventional alarm-symptom triage remains underexplored. Case presentation: A 72-year-old woman presented for outpatient evaluation with several months of pruriginous, pustular, and crusted symmetric eruption involving the dorsal aspects of the limbs, refractory to standard dermatologic treatment, and without gastrointestinal symptoms. A non-invasive systemic stool-based work-up demonstrated detectable faecal haemoglobin (iFOBT), mildly elevated faecal calprotectin (51.6 mg/kg, ULN 50 mg/kg), markedly elevated faecal alpha-1-antitrypsin (631 µg/mL; 2.3× ULN), and predominance of Escherichia coli on stool culture. Colonoscopy revealed a locally advanced rectal adenocarcinoma; staging classified the lesion as cT3N1M0. The patient received long-course neoadjuvant chemoradiotherapy (50 Gy, concurrent capecitabine) followed by low anterior resection with total mesorectal excision and pathological complete response (ypT0N0, R0), and adjuvant capecitabine. The cutaneous eruption resolved progressively in parallel with antineoplastic therapy without specific dermatologic intervention. The patient remains in remission at over 36 months. Conclusions: Persistent, unexplained, treatment-resistant pruriginous/pustular cutaneous eruptions may, in selected patients, coincide with an underlying malignancy, including colorectal cancer, and should prompt careful individualised clinical assessment, including review of age-appropriate colorectal cancer screening status. This single case raises the hypothesis that quantitative faecal immunochemical testing (FIT) may be prospectively evaluated as a low-cost, non-invasive triage tool in carefully selected patients aged ≥50 years with persistent dermatoses of unclear aetiology, even in the absence of gastrointestinal symptoms. Positive FIT results should be managed according to established local colorectal referral pathways. NICE diagnostics guidance DG56 supports FIT use in symptomatic adults with suspected lower gastrointestinal pathology; however, any extension of FIT to extraintestinal presentations remains investigational and requires formal validation through prospective studies assessing diagnostic yield, cost-effectiveness, and stage distribution. Full article

(This article belongs to the Section Clinical Diagnosis and Prognosis)

15 pages, 1083 KB

Open AccessArticle

Consensus-Level and Cluster-Adjusted Evaluation of a Large Language Model for Diagnostic Extraction from Musculoskeletal Radiology Reports

by Wolfram A. Bosbach, Elham Montazeri, Jan F. Senge, Claus Beisbart, Milena Mitrakovic, Suzanne E. Anderson, Eugen Divjak, Gordana Ivanac, Thomas Grieser, Marc-André Weber, Hatice Tuba Sanal and Keivan Daneshvar

Diagnostics 2026, 16(11), 1590; https://doi.org/10.3390/diagnostics16111590 (registering DOI) - 22 May 2026

Abstract

Purpose: Large language models (LLMs) may reduce administrative workload in radiology by automating structured diagnostic extraction from text reports. This study evaluates the accuracy of ChatGPT-4.0 when extracting correct diagnoses from musculoskeletal (MSK) radiology text reports, and compares its performance with that of [...] Read more.

Purpose: Large language models (LLMs) may reduce administrative workload in radiology by automating structured diagnostic extraction from text reports. This study evaluates the accuracy of ChatGPT-4.0 when extracting correct diagnoses from musculoskeletal (MSK) radiology text reports, and compares its performance with that of experienced human readers, using cluster-adjusted and consensus-level analyses. Materials and Methods: Twenty-three multimodal MSK imaging cases (X-ray, ultrasound, CT, and MRI) were analysed. Ten human readers and ChatGPT-4.0 (10 independent iterations) provided primary (1st) and secondary (2nd) diagnoses from six predefined options. We analysed data at the individual-reader level using cluster-adjusted generalised estimating equations (GEE) and at the case level using majority consensus with exact McNemar testing. Within-case (α_case) and within-reader (α_reader) correlations and design effects were calculated to assess clustering and implications for sample size. Results: For 1st diagnoses, AI accuracy was 0.957 (95%–CI 0.922–0.976) versus 0.865 (95%–CI 0.815–0.903) for human readers (absolute difference −0.091; OR 3.43, 95%–CI 1.07–11.02; p = 0.038). Within-case correlation (α case = 0.247) exceeded within-reader correlation (α reader ≈ 0); this resulted in a design effect of 5.7 and an effective sample size of 80.7. At the consensus level, discordance occurred in 2/23 cases (8.7%), with no significant difference between methods (McNemar p = 1.00). When 1st and 2nd diagnoses were combined, both systems achieved 23/23 correct consensus classifications. Interrater reliability between AI and human classifications was almost perfect (Gwet’s AC1 = 0.836–0.927). Conclusion and Key points: In this structured MSK text-report setting, ChatGPT-4.0 achieved diagnostic accuracy comparable to that of experienced radiologists, with modest individual-reader advantages that disappeared under consensus aggregation. Clustering analysis indicates that variability is primarily case-driven, suggesting that future validation studies will benefit more from expanding case numbers than reader numbers. Our data suggest that large performance divergences between AI and human consensus are unlikely in similar structured diagnostic contexts. Full article

(This article belongs to the Section Machine Learning and Artificial Intelligence in Diagnostics)

13 pages, 22644 KB

Open AccessArticle

Clinical Utility of AI-Enabled CT-to-MRI Translation for Degenerative Spinal Disorders: A Retrospective Reader Study

by Taehwan Kim, Hanul Gong, Hyung-Youl Park and Yeo Dong Yoon

Diagnostics 2026, 16(11), 1589; https://doi.org/10.3390/diagnostics16111589 (registering DOI) - 22 May 2026

Abstract

Background/Objectives: MRI is preferred for disc-related assessment in suspected degenerative spinal disorders, but it may be delayed, unavailable, or contraindicated; in such cases, CT findings often guide initial decisions. To address these limitations in MRI accessibility, we developed Dr.Magic (DRM-S-01), an AI-enabled [...] Read more.

Background/Objectives: MRI is preferred for disc-related assessment in suspected degenerative spinal disorders, but it may be delayed, unavailable, or contraindicated; in such cases, CT findings often guide initial decisions. To address these limitations in MRI accessibility, we developed Dr.Magic (DRM-S-01), an AI-enabled CT-to-MRI translation system that converts non-contrast spine CT into MRI-like translated images (

\hat{T}

-MRI), and conducted a retrospective reader study to assess its clinical utility. Methods: Ninety-two paired CT/MRI examinations were independently reviewed by three board-certified radiologists under three conditions: MRI-only, CT-only, and CT augmented with

\hat{T}

-MRI. MRI-only interpretation served as the per-reader reference standard. Results: CT augmented with

\hat{T}

-MRI improved diagnostic accuracy for disc-related assessment versus CT-only for all readers (34.78% to 72.83%, 42.39% to 77.17%, and 40.22% to 69.57%; all p < 0.01), increasing mean accuracy from 39.13% ± 3.20% to 73.19% ± 3.12%. Inter-reader agreement also improved (Fleiss’

κ

: 0.5617 to 0.6621; observed agreement: 0.6630 to 0.8116). Conclusions: Overall, these findings suggest that augmenting CT interpretation with

\hat{T}

-MRI may improve diagnostic performance and reading consistency when timely MRI is not feasible. In our implementation, Dr.Magic completed

\hat{T}

-MRI translation in a median of 10.90 s per CT examination (IQR, 10.39–11.79), supporting practical use within CT-based workflows. Full article

(This article belongs to the Special Issue AI for Medical Diagnosis: From Algorithms to Clinical Integration)

22 pages, 1543 KB

Open AccessArticle

Bridging Annotation Gaps: Hierarchical Self-Support Learning for Brain Tumor Segmentation

by Saqib Qamar, Mohd Fazil and Zubair Ashraf

Diagnostics 2026, 16(11), 1588; https://doi.org/10.3390/diagnostics16111588 (registering DOI) - 22 May 2026

Abstract

Background: Accurate brain tumor segmentation from Magnetic Resonance Imaging (MRI) depends on the fusion of multiple complementary modalities. However, clinical practice often faces incomplete modality sets due to acquisition failures, patient contraindications, or protocol variations. Current methods either treat each modality feature extractor [...] Read more.

Background: Accurate brain tumor segmentation from Magnetic Resonance Imaging (MRI) depends on the fusion of multiple complementary modalities. However, clinical practice often faces incomplete modality sets due to acquisition failures, patient contraindications, or protocol variations. Current methods either treat each modality feature extractor in isolation or depend on computationally expensive teacher networks for cross-modal knowledge transfer. Objective: This paper presents Hierarchical Adaptive Group Self-Support Learning with Boundary-Aware Calibration (HAGSS), a framework that overcomes three key limitations of existing group self-support methods: static group formation that ignores temporal prediction quality, uniform treatment of boundary and interior voxels, and distribution mismatch across heterogeneous modality logits. Methods: We propose a hierarchical adaptive group formation mechanism that reassigns group leader roles at each epoch based on voxel-level prediction confidence scores instead of fixed sensitivity priors. We also introduce a boundary-aware calibration module that applies spatially varied distillation weights with greater emphasis on tumor boundary regions. In addition, we design a cross-scale consistency regularization term that enforces agreement between multi-resolution predictions to stabilize the self-support target. Results: Experiments on BraTS2020, BraTS2018, and BraTS2021 datasets show that HAGSS achieves consistent improvements over state-of-the-art baselines. The average Dice gains across the whole tumor, tumor core, and enhancing tumor regions reach 1.30% on BraTS2020 and 1.61% on BraTS2021 compared to existing methods. All improvements are statistically significant (

p < 0.05

). Conclusions: HAGSS operates exclusively during training, adds no parameters or inference cost, and can be applied as a plug-in module to any multi-encoder incomplete multi-modal segmentation architecture. Code is publicly available at GitHub. Full article

(This article belongs to the Special Issue 3rd Edition: AI/ML-Based Medical Image Processing and Analysis)

15 pages, 632 KB

Open AccessArticle

Doppler Ultrasound Findings in Filler-Related Facial Vascular Adverse Events: An International Multicenter Study

by Rosa M. S. Sigrist, Claudia Gonzalez, Leonie Schelke, Ximena Wortsman, Stella Desyatnikova, Fernanda A. Cavallieri and Maria Cristina Chammas

Diagnostics 2026, 16(11), 1587; https://doi.org/10.3390/diagnostics16111587 (registering DOI) - 22 May 2026

Abstract

Background: Vascular adverse events (VAEs) related to facial filler injections are rare but potentially severe complications. Doppler ultrasound has emerged as an adjunct imaging tool for evaluating vascular compromise; however, Doppler findings in facial VAEs remain insufficiently characterized. Objectives: To characterize Doppler [...] Read more.

Background: Vascular adverse events (VAEs) related to facial filler injections are rare but potentially severe complications. Doppler ultrasound has emerged as an adjunct imaging tool for evaluating vascular compromise; however, Doppler findings in facial VAEs remain insufficiently characterized. Objectives: To characterize Doppler ultrasound findings associated with filler-related facial VAEs and to assess whether Doppler patterns differ according to prior hyaluronidase administration. Methods: This international multicenter retrospective observational study included 100 patients with clinically diagnosed facial VAEs following filler injections between May 2022 and April 2025. Doppler ultrasound findings were analyzed, including absent flow in perforators and major arteries, compensatory flow, abnormal waveforms, increased peak systolic velocity (PSV), and absence of Doppler abnormalities. Patients were categorized according to hyaluronidase administration prior to ultrasound evaluation. Descriptive statistics and comparative analyses were performed. Results: One hundred patients (median age, 38 years; IQR: 30–50; 88 women) were evaluated. The most frequent Doppler ultrasound findings were absent flow in perforators (42%) and major arteries (35%), followed by compensatory flow (26%), string sign (18%), flow diversion (16%), and increased peak systolic velocity (16%). No Doppler abnormalities were observed in 12% of cases, while tardus–parvus (9%) and staccato waveform (8%) were less frequent. Doppler ultrasound findings did not differ significantly between patients who received hyaluronidase before imaging and those who did not (all p > 0.05). The dose of hyaluronidase varied substantially. Livedo reticularis, blanching, and pain were the most common clinical findings. Central facial arterial territories, particularly the perioral, nasolabial fold, nasal, and glabellar regions, were most commonly involved. Conclusions: Filler-related facial VAEs show recognizable Doppler ultrasound patterns, and the identification of these patterns may improve localization of vascular occlusion and support ultrasound-guided hyaluronidase administration, potentially enabling more targeted delivery with lower doses. Full article

(This article belongs to the Section Clinical Diagnosis and Prognosis)

14 pages, 310 KB

Open AccessArticle

Correlations Between Ultrasound Features and Histological Findings in Adenomyosis: A Single-Center Retrospective Study

by Melinda-Ildiko Mitranovici, Dan Costachescu, Septimiu Voidazan, Liviu Moraru, Laura Caravia, Florin Bobirca, Mihai Munteanu and Romeo Micu

Diagnostics 2026, 16(11), 1586; https://doi.org/10.3390/diagnostics16111586 - 22 May 2026

Abstract

Adenomyosis is a benign gynecologic condition characterized by ectopic endometrial glands and stroma present within the myometrium. Background/Objectives: The gold standard in diagnosis is the histology of hysterectomy specimens. Due to the heterogeneity of this disease, there is a lack of valid classification. [...] Read more.

Adenomyosis is a benign gynecologic condition characterized by ectopic endometrial glands and stroma present within the myometrium. Background/Objectives: The gold standard in diagnosis is the histology of hysterectomy specimens. Due to the heterogeneity of this disease, there is a lack of valid classification. The most important symptoms are chronic pelvic pain and abnormal uterine bleeding, followed by infertility. Noninvasive diagnostic tools have been sought, with ultrasound being a valuable option. The objective of our study was to evaluate the correlation of transvaginal ultrasound, used in addition to three-dimensional ultrasonography and Doppler, with the histology of adenomyosis. Methods: An observational retrospective study was conducted between January 2015 and November 2018 on 160 women with adenomyosis managed by hysterectomy. All patients underwent transvaginal sonography combined with 3D and Doppler sonography. Results: Comparing the location of adenomyosis in the myometrium observed using ultrasound with histological findings, a statistically significant correlation was observed (p = 0.0001). Symptoms were associated with the location of the lesions, heavy period in internal adenomyosis (p ≤ 0.001), and infertility (p = 0.001), while pelvic pain was observed in external adenomyosis (p = 0.03). Deep endometriosis was associated with external adenomyosis (p = 0.001). An ill-defined junctional zone was observed via Doppler investigation in internal adenomyosis (p = 0.0001), also correlated with the symptoms. Histology confirmed all cases of adenomyosis, with statistically significant similarities regarding pattern, location, and depth (p < 0.001). Conclusions: The increasing use of 3D and Doppler evaluations enhances TVUS importance, and 3D TVUS offers high diagnostic capacity for adenomyosis, in accordance with histological findings. This procedure facilitates the adoption of therapeutic modalities other than surgery with uterus preservation. Full article

(This article belongs to the Special Issue Diagnosis and Management of Gynecological Disorders)

52 pages, 2955 KB

Open AccessReview

Near-Infrared Spectroscopy in the Pathophysiology, Diagnosis, and Exercise-Based Management of Muscle Oxygenation Impairment

by Junyan Liu, Nicolas C. Kelhofer, Tyler S. Burtner, W. Catherine Cheung, Manuel E. Hernandez and Yih-Kuen Jan

Diagnostics 2026, 16(11), 1585; https://doi.org/10.3390/diagnostics16111585 - 22 May 2026

Abstract

Muscle oxygen nation impairment, defined as a mismatch between oxygen delivery, distribution, and oxidative utilization in active skeletal muscle, contributes to exercise intolerance and functional decline. Near-infrared spectroscopy (NIRS) has emerged as the leading non-invasive tool for monitoring local muscle oxygenation, but its [...] Read more.

Muscle oxygen nation impairment, defined as a mismatch between oxygen delivery, distribution, and oxidative utilization in active skeletal muscle, contributes to exercise intolerance and functional decline. Near-infrared spectroscopy (NIRS) has emerged as the leading non-invasive tool for monitoring local muscle oxygenation, but its clinical translation and optimal exercise-based management remain incompletely defined. This scoping review aimed to (1) synthesize the pathophysiology of muscle oxygenation impairment across the oxygen transport cascade, (2) evaluate NIRS-based diagnostic protocols, and (3) review exercise-based interventions targeting muscle oxygenation. The review followed PRISMA-ScR guidelines and was prospectively registered in OSF (DOI: 10.17605/OSF.IO/QW8R3) and PROSPERO (CRD420261365040). PubMed, Web of Science, Scopus, Cochrane CENTRAL, EMBASE, PEDro, and ClinicalTrials.gov were searched through to April 2026. Methodological quality was appraised using the PEDro scale, the Downs and Black checklist, and the Newcastle–Ottawa Scale. A total of 61 studies (2003–2025) were included, with fair-to-good methodological quality (PEDro 3–8, mean 5.3; Downs and Black 15–24, mean 18.6; Newcastle–Ottawa 5–8, mean 6.5). Regarding pathophysiology, muscle oxygenation impairment is a cascade-level phenomenon with four mechanistically distinct phenotypes corresponding to the dominant site of impairment, each with characteristic NIRS signatures. Regarding diagnostic assessment, NIRS has shown value in selected contexts including a validated threshold for peripheral artery disease, but most studies report group-level correlations without deriving receiver operating characteristic curves at validated thresholds, which together with device and calibration heterogeneity limits clinical translation. Regarding exercise-based interventions, adaptations align with the underlying cascade lesion, sprint and high-intensity interval training enhance oxidative capacity, while walking-based and vascular-targeted programs preferentially improve microvascular function. These findings support a unifying framework in which the site of cascade impairment guides diagnostic protocol selection and exercise prescription. The proposed cascade lesion phenotyping schema is hypothesis-generating and requires prospective validation. Full article

(This article belongs to the Section Biomedical Optics)

24 pages, 2887 KB

Open AccessArticle

Temporal External Validation of a Customized Fetal Body Mass Index Percentile Model for Neonatal Nutritional Status Assessment

by Juan Jesús Fernández Alba, María Castillo Lara, Laura Gutiérrez Palomino, José Castro Peñas, Rocío Quintero Prado and Carmen González Macías

Diagnostics 2026, 16(11), 1584; https://doi.org/10.3390/diagnostics16111584 - 22 May 2026

Abstract

Background/Objectives: Accurate identification of neonatal malnutrition is essential for optimizing perinatal care and reducing adverse outcomes. Traditional birthweight-based methods fail to account for body proportionality, limiting their ability to distinguish constitutionally small or large neonates from those with true nutritional abnormalities. We [...] Read more.

Background/Objectives: Accurate identification of neonatal malnutrition is essential for optimizing perinatal care and reducing adverse outcomes. Traditional birthweight-based methods fail to account for body proportionality, limiting their ability to distinguish constitutionally small or large neonates from those with true nutritional abnormalities. We previously developed a customized fetal body mass index (cFBMI) percentile model that incorporates both weight and length, adjusted for maternal and fetal characteristics. This study aims to perform a temporal external validation of the cFBMI model following the Riley et al. framework, comparing its performance against the GROW customized birthweight model and the INTERGROWTH-21st population-based standard. Methods: A temporal validation study was conducted using singleton deliveries from Hospital Universitario de Puerto Real, Cádiz, Spain. The development cohort comprised 7864 deliveries (2002–2021); the validation cohort comprised 4441 deliveries (2022–2025). Inclusion criteria: singleton pregnancy, gestational age of 33–42 + 6 weeks, birthweight of 500–6000 g, known neonatal sex and length, and complete maternal data. The Ponderal Index (PI = weight/length³ × 100) stratified by sex and gestational age served as the gold standard (undernutrition: PI < p10; overnutrition: PI > p90). Discrimination was assessed using the area under the receiver operating characteristic curve (AUC) with bootstrap 95% confidence intervals (2000 iterations) and DeLong tests. Calibration was evaluated by comparing observed versus expected proportions across percentile categories. Clinical utility was assessed using decision curve analysis (DCA). Temporal stability was quantified by comparing AUCs and Brier scores between the development and validation cohorts. Results: In the validation cohort (n = 4441), cFBMI demonstrated superior discrimination for both undernutrition (AUC: 0.962) and overnutrition (AUC: 0.961) compared with GROW (AUC: 0.751 and 0.676, respectively) and INTERGROWTH-21st (AUC: 0.756 and 0.682, respectively); all DeLong comparisons p < 0.0001. The cFBMI exhibited excellent temporal stability (ΔAUC = −0.004 for undernutrition, +0.002 for overnutrition) and superior calibration (observed proportions: 9.6%/81.7%/8.8% vs. expected 10%/80%/10%; χ² = 9.22, p = 0.010). The decision curve analysis confirmed the superior net benefit of cFBMI across all threshold probabilities. Conclusions: The customized fetal BMI percentile model demonstrates excellent and temporally stable discriminative performance in this single-institution temporal validation study, with superior calibration and apparent advantages in clinical utility as determined by decision curve analysis compared with existing methods. Its integration of body proportionality provides conceptual alignment with the Ponderal Index gold standard. These findings are promising but require confirmation through external multicenter validation before clinical implementation can be recommended. Although the mathematical relationship between the index test (weight/length²) and the reference standard (weight/length³) should be considered when interpreting the magnitude of discrimination metrics, validation against independent clinical outcomes is an essential next step. The cFBMI thus provides a proportionality-aware nutritional metric whose primary discriminative advantage over weight-based methods is realized at and beyond the moment of birth, and which is forward-compatible with emerging modalities for independent prenatal fetal length estimation. Full article

(This article belongs to the Section Clinical Diagnosis and Prognosis)

► Show Figures

Graphical abstract

21 pages, 1041 KB

Open AccessArticle

Deep Feature–Based Detection of Chiari Malformation Type I from Sagittal T2-Weighted MRI Using a Hybrid CNN–Machine Learning Framework

by Zülküf Akdemir and Murat Canayaz

Diagnostics 2026, 16(11), 1583; https://doi.org/10.3390/diagnostics16111583 - 22 May 2026

Abstract

Objective: Chiari Type I Malformation (CM1) is a structural abnormality of the hindbrain that can cause a range of neurological symptoms and often requires radiological confirmation using magnetic resonance imaging (MRI). The aim of this study was to develop and evaluate a deep [...] Read more.

Objective: Chiari Type I Malformation (CM1) is a structural abnormality of the hindbrain that can cause a range of neurological symptoms and often requires radiological confirmation using magnetic resonance imaging (MRI). The aim of this study was to develop and evaluate a deep feature-based machine learning framework for the automated detection of CM1 from sagittal MRI images. Materials and Methods: The cohort comprised 550 adults: 250 patients with CM1 (168 women, 82 men; age range, 18–65 years) and 300 healthy control participants (210 women, 90 men; age range, 18–65 years). A total of 764 T2-weighted sagittal MR images (384 CM1, 380 healthy) acquired from two different 1.5T MRI scanners (Siemens Magnetom Altea and Symphony) between 2020 and 2024 were retrospectively analyzed. Deep features were extracted using ResNet-50 and MobileNetV2 architectures and subsequently classified using Support Vector Machines (SVM), Logistic Regression (LR), Random Forest (RF), XGBoost, and voting-based ensemble models. Model performance was assessed through patient-level 5-fold cross-validation using accuracy, sensitivity, specificity, F1-score, PPV, NPV, and AUC metrics. Code and trained models are available from the corresponding author upon reasonable request; imaging data are not publicly available due to patient privacy and institutional restrictions. Results: Across patient-level five-fold cross-validation, models built on ResNet-50 deep features demonstrated extremely high and stable diagnostic performance. The final soft-voting ensemble classifier based on ResNet-50 achieved perfect mean performance, with accuracy, balanced accuracy, sensitivity, specificity, F1-score, and AUC all equal to 1.000 ± 0.000 across folds. Other ResNet-based classifiers also achieved near-perfect results. MobileNetV2-based models also demonstrated strong performance but showed slightly lower stability compared with ResNet-based models, with mean accuracies ranging from 0.984 to 0.993 and mean AUC values between 0.99947 and 0.99984 across classifiers. Conclusion: The proposed deep feature-based machine learning framework demonstrated excellent performance for the automated detection of Chiari Type I Malformation from sagittal MRI images. In particular, the ResNet-50–based soft-voting ensemble model achieved perfect classification performance in cross-validation testing, suggesting that deep feature representations combined with machine learning classifiers may serve as a promising computer-aided diagnostic tool for supporting radiological evaluation of CM1. Full article

(This article belongs to the Section Machine Learning and Artificial Intelligence in Diagnostics)

27 pages, 1158 KB

Open AccessReview

Microbiomics: Novel Biomarkers of Colorectal Cancer Diagnosis and Prognosis

by Lielong Yang, Wenjian Meng, Tinghan Yang, Yuzhou Zhu and Ziqiang Wang

Diagnostics 2026, 16(11), 1582; https://doi.org/10.3390/diagnostics16111582 - 22 May 2026

Abstract

With colorectal cancer (CRC) accounting for over 1.9 million new cases and 930,000 deaths globally in 2020, there is a critical need for innovative indicators to forecast disease advancement and therapeutic outcomes. The gut microbiome has emerged as a fertile area for discovering [...] Read more.

With colorectal cancer (CRC) accounting for over 1.9 million new cases and 930,000 deaths globally in 2020, there is a critical need for innovative indicators to forecast disease advancement and therapeutic outcomes. The gut microbiome has emerged as a fertile area for discovering such diagnostic and prognostic signals. This narrative review collected current evidence on intestinal microorganisms and their metabolic products as candidate markers for CRC control. Intestinal communities influence malignancy through diverse mechanisms, including metabolic shifts, immune modulation, inflammation, proliferation/apoptosis regulation, genotoxicity, and mucosal barrier disruption. Pathogenic species, such as Fusobacterium nucleatum and enterotoxigenic Bacteroides fragilis, facilitate tumorigenesis via FadA-mediated signaling and Th17/IL-17 responses. In contrast, beneficial taxa like Faecalibacterium prausnitzii and Akkermansia muciniphila provide protective effects through short chain fatty acid production. Macrophage phenotype physiological equilibrium is altered and inflammatory status fluctuates under the former. Metabolically, hydrogen sulfide damages mitochondrial DNA and secondary bile acids stimulate cellular proliferation. While 16S rRNA sequencing and shotgun metagenomics are established detection strategies, innovative platforms like organoids and gene arrays remain in the exploratory stage. Clinical data indicates that F. nucleatum aligns with advanced tumor stage, and its combined detection with colibactin-producing E. coli achieves high sensitivity for early-stage screening. Additionally, A. muciniphila levels can anticipate the efficacy of PD-1 blockade immunotherapy. Microbiota-derived tools represent a transformative direction in oncology. Future research must focus on standardizing protocols and validating multi-marker panels to enhance clinical translation. Full article

(This article belongs to the Special Issue Recent Advances in Microbiome-Based Diagnostics for Oncological Diseases: Innovations and Clinical Applications)

► Show Figures

Figure 1

14 pages, 3508 KB

Open AccessReview

Advanced Preoperative Imaging in Macula-Off Rhegmatogenous Retinal Detachment: Emerging Diagnostic and Prognostic Insights for Clinical Management

by Lorenzo Motta, Rodolfo Mastropasqua, Michele Cillis, Giulia Craighero, Nicola Sereni, Corina De Santis, Alberto Quarta, Aldo Gelso, Giuseppe Lo Giudice and Claudio Iovino

Diagnostics 2026, 16(11), 1581; https://doi.org/10.3390/diagnostics16111581 - 22 May 2026

Abstract

Retinal detachment (RD) is a potentially sight-threatening condition that requires timely diagnosis and appropriate surgical management. In macula-off rhegmatogenous retinal detachment (RRD), visual recovery after successful reattachment remains highly variable, highlighting the need for reliable preoperative prognostic markers. This study focuses on the [...] Read more.

Retinal detachment (RD) is a potentially sight-threatening condition that requires timely diagnosis and appropriate surgical management. In macula-off rhegmatogenous retinal detachment (RRD), visual recovery after successful reattachment remains highly variable, highlighting the need for reliable preoperative prognostic markers. This study focuses on the contribution of advanced retinal imaging to the preoperative assessment of macula-off RRD, summarizing current evidence on imaging-derived biomarkers associated with disease severity and postoperative functional outcome. In this narrative review, we analyze studies employing spectral-domain and swept-source optical coherence tomography (SD-OCT and SS-OCT), OCT angiography (OCT-A), and adaptive optics OCT (AO-OCT) to characterize microstructural and microvascular retinal alterations. Emerging approaches, including ultra-widefield OCT (UWF-OCT) and artificial intelligence-based image analysis, are also discussed for their potential role in refining diagnosis, supporting surgical planning, and improving prognostic stratification. While several imaging parameters appear promising, their prognostic value is not yet fully realized. Further prospective studies are required to validate clinically meaningful imaging biomarkers and to integrate advanced imaging into routine preoperative decision-making for macula-off RRD. Full article

(This article belongs to the Special Issue Applications of Optical Coherence Tomography in the Ocular Diagnosis: From the Tear Film to the Sclera 2.0)

► Show Figures

Figure 1

19 pages, 654 KB

Open AccessReview

Pediatric Desmoplastic Fibroma of the Jaws: A Comprehensive Review of Clinical Presentation, Management, and Outcomes

by George Batshon, Murad Abdelraziq, Imad Abu El-Naaj and Yasmine Ghantous

Diagnostics 2026, 16(11), 1580; https://doi.org/10.3390/diagnostics16111580 - 22 May 2026

Abstract

Background: Desmoplastic fibroma (DF) is a rare, benign, but locally aggressive intraosseous tumor with a predilection for the mandible in pediatric patients. Owing to its low incidence, evidence guiding management remains limited. Objective: To provide a comprehensive review of the clinical presentation, radiographic [...] Read more.

Background: Desmoplastic fibroma (DF) is a rare, benign, but locally aggressive intraosseous tumor with a predilection for the mandible in pediatric patients. Owing to its low incidence, evidence guiding management remains limited. Objective: To provide a comprehensive review of the clinical presentation, radiographic features, treatment strategies, and outcomes of pediatric DF of the jaws. Methods: A comprehensive literature review was conducted using PubMed/MEDLINE, Embase, Cochrane Library, and IEEE Xplore to identify relevant studies published between 2000 and 2026. Given the rarity of this entity, a broad search strategy was applied. Eligible studies were analyzed to extract data on patient demographics, clinical features, imaging findings, treatment modalities, and outcomes. Results: A total of 32 studies comprising 45 pediatric cases were identified. The mandible was involved in 86.7% of cases. The most common presentation was painless swelling or facial asymmetry (68.9%). Wide or segmental resection was the primary treatment in 68.9% of cases. Recurrence data were available for 75.6% of cases, with an overall recurrence rate of 2.9%, occurring following incomplete resection. Conclusions: Pediatric DF of the jaws is a rare but locally aggressive tumor requiring accurate diagnosis and individualized surgical management. Complete resection with clear margins appears to provide the most reliable outcomes. However, interpretation of outcomes is limited by the predominance of case reports, heterogeneous reporting, and incomplete follow-up. Future multicenter studies and standardized reporting are needed to better define optimal management strategies. Full article

(This article belongs to the Special Issue Diagnosis and Management in Oral and Maxillofacial Surgery)

► Show Figures

Figure 1

15 pages, 1929 KB

Open AccessArticle

Prediction of Surgical Intervention in Acute Knee Trauma: A Focus on Threshold-Specific Performance and Clinical Decision Utility

by Eun Byeol Choe, Joungeun Lee, Won-Kee Choi, Young Woo Seo and Sang Gyu Kwak

Diagnostics 2026, 16(11), 1578; https://doi.org/10.3390/diagnostics16111578 - 22 May 2026

Abstract

Background: Acute knee trauma is a common reason for emergency department visits, yet early identification of patients requiring surgical intervention remains challenging. Most existing prediction studies focus on discrimination metrics and provide limited guidance for clinical decision-making. Methods: We conducted a [...] Read more.

Background: Acute knee trauma is a common reason for emergency department visits, yet early identification of patients requiring surgical intervention remains challenging. Most existing prediction studies focus on discrimination metrics and provide limited guidance for clinical decision-making. Methods: We conducted a retrospective study of 905 patients presenting to the emergency department with acute knee trauma. Prediction models were developed using logistic regression, random forest, and extreme gradient boosting (XGBoost) based on routinely available clinical variables. Model performance was evaluated in terms of discrimination (AUROC, AUPRC), calibration, and clinical utility. Threshold-specific performance metrics and decision curve analysis were used to assess clinical applicability, and patients were stratified into risk groups based on predicted probabilities. Results: Among 905 patients, 163 (18.0%) underwent surgical intervention. Logistic regression and random forest demonstrated comparable performance (AUROC 0.748 and 0.744, respectively), whereas XGBoost showed lower discrimination (AUROC 0.632). Calibration was acceptable overall but less stable at higher predicted probabilities. Threshold-specific analysis demonstrated meaningful trade-offs between sensitivity and specificity across probability thresholds. Decision curve analysis showed that the model provided greater net benefit than default strategies within a threshold range of approximately 0.05–0.25. Risk stratification showed increasing surgical rates across risk groups, although the degree of separation was modest. Conclusions: Prediction models based on routinely available clinical variables can support early risk assessment in acute knee trauma. Their clinical usefulness depends on threshold-specific evaluation and decision-analytic approaches rather than overall performance metrics alone. These findings highlight the importance of interpreting prediction models within a clinical decision-making framework to facilitate real-world application. Full article

(This article belongs to the Special Issue Advances in Disease Prediction—2nd Edition)

► Show Figures

Figure 1

11 pages, 780 KB

Open AccessArticle

Lung Auscultation for Detecting Interstitial Lung Disease in Patients with Newly Diagnosed Systemic Sclerosis: Retrospective Cohort Study

by Felix W. Wireko, Vasilios Tzilas, Comfort Anim-Koranteng, Ahmed S. Sayed Ahmed, Yvette A. Yeboah-Kordieh, Ashima Makol and Jay H. Ryu

Diagnostics 2026, 16(11), 1577; https://doi.org/10.3390/diagnostics16111577 - 22 May 2026

Abstract

Background/Objectives: Interstitial lung disease (ILD) occurs commonly in systemic sclerosis (SSc) and is the leading cause of mortality. There are limited data on the accuracy of lung auscultation in identifying the presence of ILD in patients with SSc. Methods: We retrospectively identified patients [...] Read more.

Background/Objectives: Interstitial lung disease (ILD) occurs commonly in systemic sclerosis (SSc) and is the leading cause of mortality. There are limited data on the accuracy of lung auscultation in identifying the presence of ILD in patients with SSc. Methods: We retrospectively identified patients with newly diagnosed SSc who had documented lung auscultation findings and chest high-resolution computed tomography (HRCT) available for review. Diagnoses were made by rheumatologists at Mayo Clinic, Rochester, Minnesota, USA over a 4-year period. Pulmonary function measurements included lung volumes, spirometry, and single-breath diffusing capacity. Results: Among 151 patients with SSc (median age, 62 years), 72.2% were female and 55.0% were never smokers. Limited cutaneous SSc was the most common phenotype (67.3%). Seventy (46.4%) patients were evaluated by pulmonologists. There was evidence of ILD by HRCT in 69 patients (45.7%); the most common pattern of ILD was fibrotic nonspecific interstitial pneumonia (59.2%). Respiratory symptoms were present in 46.4% of those with ILD compared to 15.9% among those without. The sensitivity and specificity for crackles heard by rheumatologists in detecting ILD were 50.7% and 97.6%, respectively; for pulmonologists, 71.4% and 85.7%, respectively. Presence of crackles was associated with high positive predictive values (94.6% for rheumatologists vs. 92.1% for pulmonologists, respectively); negative predictive values were moderate (70.2% vs. 56.3%, respectively). Crackles correlated with lower pulmonary function measures but did not differ across ILD patterns. Conclusions: Detection of crackles on lung auscultation appears to be a specific and moderately sensitive indicator of ILD (often asymptomatic) in patients with newly diagnosed SSc. The presence of crackles correlates with worse pulmonary function but may be absent in early ILD. Full article

(This article belongs to the Section Clinical Diagnosis and Prognosis)

► Show Figures

Figure 1

31 pages, 4069 KB

Open AccessReview

Tuberculosis in Pregnancy: An Updated Narrative Review

by Carolina Longo, Karina Felippe Monezi Pontes, Marina Matos de Moura Faíco, Mayra Martins Melo, Gustavo Yano Callado, Célio de Barros Barbosa, Edward Araujo Júnior and Antonio Braga

Diagnostics 2026, 16(11), 1576; https://doi.org/10.3390/diagnostics16111576 - 22 May 2026

Abstract

Tuberculosis remains one of the leading infectious causes of morbidity and mortality worldwide, disproportionately affecting women of reproductive age, particularly in low- and middle-income countries. Tuberculosis during pregnancy represents a major clinical challenge, as physiological and immunological changes associated with pregnancy may obscure [...] Read more.

Tuberculosis remains one of the leading infectious causes of morbidity and mortality worldwide, disproportionately affecting women of reproductive age, particularly in low- and middle-income countries. Tuberculosis during pregnancy represents a major clinical challenge, as physiological and immunological changes associated with pregnancy may obscure symptoms, delay diagnosis, and contribute to adverse maternal and perinatal outcomes. This narrative review provides an updated and clinically oriented overview of tuberculosis during pregnancy, with particular emphasis on diagnostic challenges, imaging strategies, microbiological testing, maternal–fetal complications, and therapeutic management. Key topics include symptom-based screening, tuberculin skin test and interferon gamma release assays, as well as molecular diagnostic methods such as GeneXpert Mycobacterium tuberculosis/Rifampicin (MTB/RIF) and Xpert MTB/RIF Ultra, chest radiography, computed tomography, and emerging biomarkers. We also discuss the impact of tuberculosis on pregnancy outcomes, including prematurity, low birth weight, maternal morbidity, and neonatal complications, as well as the particular challenges posed by human immunodeficiency virus HIV coinfection and multidrug-resistant tuberculosis. Current treatment strategies, preventive approaches, postpartum care, neonatal management, and Bacille Calmette–Guérin vaccination are reviewed in light of contemporary evidence and international recommendations. Finally, we highlight practical diagnostic algorithms, current evidence gaps, and priorities for future research aimed at improving maternal and neonatal outcomes in both high- and low-resource settings. Full article

(This article belongs to the Section Diagnostic Microbiology and Infectious Disease)

► Show Figures

Figure 1

4 pages, 943 KB

Open AccessInteresting Images

A Giant Exophytic Gastric GIST Mimicking Ovarian Cancer: A Diagnostic Pitfall on CT and [¹⁸F]FDG PET/CT

by Sang Jun Byun, Sun-Jae Lee and Byungwook Choi

Diagnostics 2026, 16(11), 1575; https://doi.org/10.3390/diagnostics16111575 - 22 May 2026

Abstract

A 66-year-old woman was referred for evaluation of a large pelvic mass suspected to be ovarian cancer. Contrast-enhanced computed tomography (CECT) revealed a giant multiseptated cystic pelvic mass with enhancing solid components; although its superior aspect closely abutted the gastric serosa, its predominant [...] Read more.

A 66-year-old woman was referred for evaluation of a large pelvic mass suspected to be ovarian cancer. Contrast-enhanced computed tomography (CECT) revealed a giant multiseptated cystic pelvic mass with enhancing solid components; although its superior aspect closely abutted the gastric serosa, its predominant pelvic location raised concern for an adnexal malignancy. Subsequent [¹⁸F]fluorodeoxyglucose positron emission tomography/computed tomography ([¹⁸F]FDG PET/CT) demonstrated mild uptake confined to the viable solid portion (SUVmax 2.72) without hypermetabolic nodal or distant metastases. Exploratory laparotomy revealed a giant pedunculated tumor arising from the gastric antrum and descending into the pelvis. Histopathology confirmed an epithelioid gastrointestinal stromal tumor positive for CD117, DOG1, and CD34. This case highlights an important diagnostic pitfall in which a giant exophytic gastric GIST may mimic ovarian cancer because of its pelvic location and cystic-solid appearance. Careful correlation of CECT, fused [¹⁸F]FDG PET/CT, and pathologic findings is essential for accurate assessment of the organ of origin in large abdominopelvic masses. Full article

(This article belongs to the Section Medical Imaging and Theranostics)

► Show Figures

Figure 1

23 pages, 9873 KB

Open AccessArticle

RNNet-MST: A ResNet-50 with Multi-Scale Transformer Blocks for Pulmonary Nodule Classification and Attention-Based Localization on Chest X-Ray Images

by Edrill F. Bilan, Emman T. Manduriaga, Hernando S. Salapare III, Ymir M. Garcia, Khatalyn E. Mata, Rose Anna R. Banal, Imelda C. Ang, Wei-Ta Chu and Dan Michael A. Cortez

Diagnostics 2026, 16(10), 1574; https://doi.org/10.3390/diagnostics16101574 - 21 May 2026

Abstract

Background/Objectives: Lung cancer survival depends on early detection; however, in the Philippines, high radiologist workloads and the anatomical complexity of chest X-rays (CXRs) contribute to missed pulmonary nodules and false-negative diagnoses. This study aims to develop an enhanced deep learning model to [...] Read more.

Background/Objectives: Lung cancer survival depends on early detection; however, in the Philippines, high radiologist workloads and the anatomical complexity of chest X-rays (CXRs) contribute to missed pulmonary nodules and false-negative diagnoses. This study aims to develop an enhanced deep learning model to improve nodule classification and localization sensitivity. Methods: We propose RNNet-MST, an extension of ResNet-50 that incorporates Multi-Scale Transformer blocks for global context modeling and a custom spatial attention mechanism for attention-based weak localization of disease-relevant regions. The model was trained and evaluated on the NODE21 chest X-ray dataset and compared with a baseline ResNet-50 using classification metrics, with attention maps used for weak localization analysis. Results: RNNet-MST demonstrated consistent improvements over the baseline ResNet-50 across evaluated metrics. Mean Nodule Recall improved from 88.02 ± 1.92% to 91.55 ± 1.41%, reducing false negatives. Mean Test Precision reached 90.46 ± 0.99%, and mean Nodule F1-Score improved to 90.99 ± 0.39%. On the isolated small-nodule subset, RNNet-MST achieved a 12.3% improvement in sensitivity over the baseline. Conclusions: The integration of multi-scale transformer features improved classification sensitivity, while the attention mechanism provided weak localization cues that aligned more closely with annotated nodule regions than the baseline. RNNet-MST shows potential as a diagnostic support tool, warranting further validation on larger and more diverse clinical datasets to reduce perceptual errors and facilitate early lung cancer detection in resource-constrained settings. Full article

(This article belongs to the Section Machine Learning and Artificial Intelligence in Diagnostics)

► Show Figures

Figure 1

32 pages, 1961 KB

Open AccessReview

When Immunophenotype Is Not Identity: A Clinicopathological Review of Neuroendocrine Differentiation in Tumors of the Female Genital Tract

by Catalin-Bogdan Satala, Alina-Mihaela Gurau, Gabriela Patrichi, Roxana-Cristina Mehedinti, Andy Radu Leibovici and Gabriela Gurau

Diagnostics 2026, 16(10), 1573; https://doi.org/10.3390/diagnostics16101573 - 21 May 2026

Abstract

Neuroendocrine differentiation in tumors of the female genital tract is an uncommon but diagnostically consequential finding. Its interpretation is challenging because neuroendocrine marker expression does not necessarily define a neuroendocrine neoplasm. Focal or aberrant staining for synaptophysin, chromogranin A, CD56 or INSM1 may [...] Read more.

Neuroendocrine differentiation in tumors of the female genital tract is an uncommon but diagnostically consequential finding. Its interpretation is challenging because neuroendocrine marker expression does not necessarily define a neuroendocrine neoplasm. Focal or aberrant staining for synaptophysin, chromogranin A, CD56 or INSM1 may occur in otherwise conventional gynecologic carcinomas, whereas true poorly differentiated neuroendocrine carcinomas represent aggressive tumors with distinct prognostic and therapeutic implications. This narrative review examines neuroendocrine differentiation across the cervix, endometrium, ovary, vagina and vulva from an integrated clinicopathologic perspective. We emphasize that neuroendocrine differentiation should be approached as a diagnostic and biological spectrum, ranging from incidental immunophenotypic expression to carcinoma with neuroendocrine differentiation, mixed neuroendocrine/non-neuroendocrine tumors, well-differentiated neuroendocrine tumors and poorly differentiated neuroendocrine carcinomas. Morphology remains the diagnostic anchor, while immunohistochemistry, molecular context and clinicoradiologic correlation refine classification and help exclude mimics or metastatic disease. Site-specific interpretation is essential: cervical neuroendocrine carcinoma is commonly HPV-associated and clinically aggressive; endometrial tumors require integration with p53, mismatch repair, POLE and SWI/SNF-related contexts; ovarian lesions demand distinction between primary well-differentiated neuroendocrine tumors, poorly differentiated carcinomas and metastases; and vaginal or vulvar tumors require careful exclusion of adjacent extension, cutaneous mimics and extragenital primaries. We propose a practical diagnostic framework that separates incidental marker expression from clinically meaningful neuroendocrine differentiation and links this distinction to reporting, prognosis and treatment. The central diagnostic question is not whether neuroendocrine markers are expressed but whether their expression defines a morphologically, biologically and clinically meaningful tumor category. Full article

(This article belongs to the Section Pathology and Molecular Diagnostics)

12 pages, 1079 KB

Open AccessArticle

Enhanced Prediction of Cardiovascular Disease Through Integrated Machine Learning Models Combining Clinical and Demographic Characteristics

by Zhe Zhang, Dengao Li, Jumin Zhao, Huiting Ma, Fei Wang and Qinglian Hao

Diagnostics 2026, 16(10), 1572; https://doi.org/10.3390/diagnostics16101572 - 21 May 2026

Abstract

Background/Objectives: Heart failure (HF) remains a major cause of global mortality and morbidity; it is, therefore, of paramount importance that diagnosis and prognostication are made timely in order to better improve outcomes and reduce healthcare expenditure. This research presents a novel predictive model [...] Read more.

Background/Objectives: Heart failure (HF) remains a major cause of global mortality and morbidity; it is, therefore, of paramount importance that diagnosis and prognostication are made timely in order to better improve outcomes and reduce healthcare expenditure. This research presents a novel predictive model of heart failure that combines clinical criteria with demographic factors in order to maximize predictive performance and act as a reliable tool for individualized healthcare intervention. Methods: Complex machine learning techniques, including decision trees, random forest, and deep learning, are applied in analyzing a large dataset of subjects with heart failure. We collected a diverse dataset comprising clinical indicators such as echocardiographic data, biomarkers, electrocardiogram (ECG) features, and demographic information. Data preprocessing techniques, such as feature normalization and handling of missing values, were applied to ensure the integrity and reliability of the dataset. Results: The results indicate that integrating both clinical indicators and demographic characteristics significantly improves the predictive power of the model, compared to models based on clinical indicators alone. Specifically, the hybrid model demonstrated a superior ability to predict short- and long-term outcomes in heart failure patients, offering enhanced accuracy in risk stratification and prognosis prediction. Conclusions: This research highlights the potential of artificial intelligence (AI) and machine learning in revolutionizing heart failure care by providing healthcare professionals with more accurate, data-driven decision support tools. The proposed model not only holds promise for clinical applications but also offers insights for future research into personalized medicine. Full article

(This article belongs to the Section Machine Learning and Artificial Intelligence in Diagnostics)

► Show Figures

Figure 1

Journal Description

Diagnostics

Latest Articles

Journal Menu

Journal Browser

Highly Accessed Articles

Latest Books

E-Mail Alert

News

Topics

Conferences

Special Issues

Topical Collections

Further Information

Guidelines

MDPI Initiatives

Follow MDPI