Veterinary Sciences

Early and accurate identification of mastitis-associated inflammation is critical to increasing animal comfort, reducing financial losses, and enhancing milk quality. Although infrared thermography (IRT) has emerged as a viable non-invasive screening technique, the relative efficacy of convolutional neural networks (CNNs) and Vision Transformers (ViTs) for automated CMT-status classification from thermal udder pictures is still unknown. This work carefully analyzed seven ImageNet-pretrained deep learning architectures using 976 thermal udder images (708 healthy and 268 mastitic images) from 488 Holstein cows (354 healthy cows, 708 images; 134 mastitic cows, 268 images), including two Vision Transformer models (ViT-B/16 and Swin-Tiny) and five CNN models (ResNet-50, DenseNet-121, EfficientNet-B0, MobileNetV2, Inception-V3). Before training, pictures were preprocessed using contrast-limited adaptive histogram equalization (CLAHE), scaled to 224 × 224 pixels, and divided using cow-oriented grouping intended to reduce animal-level data leakage (this grouping could not be independently verified against a ground-truth cow roster; see Limitations). Each model was developed from start to finish and evaluated using an independent hold-out test set and five-fold animal-level cross-validation. DenseNet-121 had the greatest results on the hold-out test set, with an accuracy of 82.2%, an AUC of 0.922, a sensitivity of 90.0%, and a specificity of 79.2%. Additionally, it achieved the highest cross-validation performance (mean AUC = 0.914 ± 0.019). According to statistical analysis, DenseNet-121 was statistically indistinguishable from Inception-V3 under both tests and from ResNet-50 under the more conservative corrected resampled t-test, but significantly outperformed EfficientNet-B0, MobileNetV2, and both Vision Transformers (p < 0.05), while all CNN models outperformed both Vision Transformer models (p < 0.05). Explainable artificial intelligence (XAI) investigation utilizing Grad-CAM corroborated the biological plausibility of the learnt plausibility characteristics by showing that heat patterns in the udder region had a significant impact on model predictions. Grad-CAM analysis showed that heat patterns in the udder region partly drove model predictions, though attention was occasionally influenced by background regions. The best-performing model was tested without retraining on an independent external cohort of 85 cows from a different farm. Although external performance decreased (accuracy = 60.0%; Cohen’s κ = 0.199), consistent with the expected effects of domain shift, the somatic cell count was significantly higher in CMT-positive cows (p < 0.001; AUC = 0.783), supporting the biological significance of the identified thermal abnormalities. Fine-tuned CNNs, led by DenseNet-121, delivered accurate and reproducible thermal CMT-status classification under internal cross-validation, but transportability to an independent farm was limited (external accuracy = 0.600, κ = 0.199), indicating that domain adaptation and multi-farm validation are needed before broader deployment. These findings provide strong support for the application of CNN-based deep learning in IRT-assisted CMT-status screening while highlighting the necessity for larger multicenter datasets to further evaluate transformer-based approaches. aptation and multi-farm validation are needed before broader deployment. These findings provide strong support for the application of CNN-based deep learning in IRT-assisted mastitis screening while highlighting the necessity for larger multicenter datasets to further evaluate transformer-based approaches.

Bovine respiratory disease (BRD) is a major cause of morbidity and mortality in cattle, and Klebsiella pneumoniae (K. pneumoniae) has emerged as an opportunistic pathogen that can colonize the bovine respiratory tract. However, epidemiological data on K. pneumoniae from cattle with BRD in Henan Province, central China, remain scarce. This study aimed to investigate the isolation rate, antimicrobial resistance profiles, resistance genes, virulence-associated genes, and genetic diversity of K. pneumoniae from BRD-affected calves in Henan. From 2021 to 2026, deep nasal swabs were collected once from 517 calves with clinical signs of BRD (Wisconsin score ≥5) across 38 farms in 13 cities over six years. A total of 87 K. pneumoniae isolates were obtained, giving an overall isolation rate of 16.8% (87/517), ranging from 13.8% to 22.6% across five geographic regions. Antimicrobial susceptibility was determined by disk diffusion using CLSI VET01S and CLSI M100-S26 criteria for 15 antimicrobials belonging to six classes. Resistance and virulence-associated genes were screened by PCR, and multilocus sequence typing (MLST) was performed. MLST revealed 34 sequence types (STs), with ST950 being the most prevalent (6.9%), followed by ST262, ST187, and ST306. Antimicrobial resistance was highest against ampicillin (94.3%), sulfisoxazole (79.3%), cefoxitin (73.6%), cefazolin (70.1%), and enrofloxacin (67.8%). Multidrug resistance (resistance to ≥3 classes) was observed in 96.6% (84/87; 95% CI 90.3–98.8%) of isolates. The most common resistance genes were blaTEM (72.4%), sul1 (39.1%), blaSHV (37.9%), and qnrB (14.9%). After Firth’s penalized logistic regression with farm-level cluster-robust standard errors and Benjamini–Hochberg FDR correction, no resistance gene was significantly associated with phenotypic resistance (all q > 0.05). Virulence-associated genes fimH, mrkD, uge, wabG, entB, iutA, and ureA were highly prevalent (>90%), whereas hypervirulence markers (rmpA and rmpA2) were absent. The most common virulence-associated gene combination (eight genes) was present in 35.6% of isolates. These findings indicate high genetic diversity, widespread multidrug resistance, and a broad virulence-associated gene repertoire among K. pneumoniae isolated from cattle with BRD in Henan. Because no healthy comparison group or etiological confirmation was included, the study does not establish K. pneumoniae as the causative agent of BRD. The high resistance rates to β-lactams, sulfonamides, and quinolones highlight the need for prudent antimicrobial use. Florfenicol showed the lowest in vitro resistance rate among the tested agents; however, clinical efficacy, treatment outcome, and safety were not evaluated. Continuous surveillance and molecular monitoring are essential to manage this pathogen in cattle populations.

Theileria annulata causes tropical theileriosis in cattle and is endemic in Kazakhstan. Although molecular diagnostic methods for this pathogen exist, data on its molecular surveillance and genetic diversity in Kazakhstan remain limited. This study aimed to develop and experimentally validate a species-specific real-time PCR (qPCR) assay for the detection of Theileria annulata and to conduct molecular surveillance and phylogenetic analysis of circulating isolates. A total of 709 cattle blood samples from 10 regions of Kazakhstan were analysed. Species-specific primers and a probe were designed targeting the 18S rRNA gene, and diagnostic performance was evaluated against sequencing as the reference method. Theileria annulata infection was confirmed in 110 of 709 samples (15.51%), with positive samples identified in the Turkistan, West Kazakhstan, Kyzylorda, and Zhambyl regions. The developed assay showed high diagnostic sensitivity (99.09%), specificity (99.67%), and diagnostic efficiency (99.58%), with no cross-reactivity with Theileria orientalis, Theileria parva, Babesia bovis, Babesia bigemina, or Anaplasma marginale. Phylogenetic analysis of partial 28S rRNA sequences confirmed the species identity of the detected isolates and demonstrated their close relationship with previously reported Kazakhstani and Chinese isolates. The developed qPCR assay is suitable for the molecular diagnosis and surveillance of bovine tropical theileriosis in Kazakhstan.