Detection and Model of Thermal Traces Left after Aggressive Behavior of Laboratory Rodents

Round 1

Reviewer 1 Report

• The manuscript is well written.
• The simulation results hard to interpret the effective of proposed method. If you have any lab results or another study to compare, that will be very support for the simulation results.
• Some figures need to be rearrange (2,3,8,12,13).

Author Response

Point 1: The simulation results hard to interpret the effective of proposed method. If you have any lab results or another study to compare, that will be very support for the simulation results.

Response 1: Thank you for your Review. 

The presented simulation tests were aimed at creating conditions for observing the possibility of detecting traces of saliva in time. Simulations of the shape or movement of animals were not the subject of this work.

Temperature simulations were based on the observations described in the paper: [25] Mazur-Milecka, M.; Ruminski, J. The analysis of temperature changes of the saliva traces left on the fur during laboratory rats social contacts. 2018 40th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) 2018, pp. 2607–2610. doi:10.1109/EMBC.2018.8512743.

Other laboratory results of trace detection are also presented in papers referenced as [18] and [32] in the manuscript.

Point 2: Some figures need to be rearrange (2,3,8,12,13).

Response 2: This format for figures is allowed. Thanks to this arrangement, the whole work is concise.

Reviewer 2 Report

Author Response

Thank you very much for your Review.

Reviewer 3 Report

Review for the paper : Detection and model of thermal traces left after aggressive behavior of laboratory rodents

This papers deals with developing a method for an analysis of identifying aggressive behavior in thermal images methodby detecting traces of saliva left on the animals’ fur after a bite, nape attack or grooming. The proposed algorithm together with thermal imaging provides additional data necessary to automate the analysis of social behavior in rodents.

The paper is well written and should be very helpful for the community of researchers that are interesting on automatic analysis behaviors of rodents in the lab. The authors are using methods and algorithms involving image processing (corner detection) and texture simulation for expanding the data for better understanding the problem.

I recommend to accept the paper with some minor changes :

1. It is not clear for me the reason for increasing the data set with simulations, usually increasing the data is a method which is commonly used for data augmentation for deep learning purposes. As for my understanding here, the purpose is different. Could you explain the motivation for the simulated shapes of the rodent and the silva. ?
2. In the Harris algorithm for corner detection, how the value of 0.000001 was determined?

Comments for author File: Comments.pdf

Author Response

Review 3
Thank you very much for your revision.

1. It is not clear for me the reason for increasing the data set with simulations, usually increasing the data is a method which is commonly used for data augmentation for deep learning purposes. As for my understanding here, the purpose is different. Could you explain the motivation for the simulated shapes of the rodent and the silva. ?

The aim of the simulation was to increase the amount of data about the bite and to control its parameters. Performing detection on such simulated images allowed for the definition of detection conditions and the selection of the best methods and parameters.
I would also like to point out that bites, resulting from social aggression, are very difficult to observe in natural images due to the high speed of the animals during an attack and the close physical contact that makes it almost impossible to observe uncovered animals.

2. In the Harris algorithm for corner detection, how the value of 0.000001 was determined?

Detection coefficients were experimentally selected, this is described in paper [32] Mazur-Milecka, M. Thermal imaging in automatic rodent’s social behaviour analysis. QIRT; ,
2016; pp. 563–569. doi:10.21611/qirt.2016.083.

Author Response File: Author Response.pdf

Reviewer 4 Report

Major:
Please add illustrative cases with discussion when algorithm fails.

Minors:
Fig. 8 quality of images is low - there are some vertical lines visible

Tab.1 \times not * operator

Eq.4 brackets size too small

Author Response

Major:
1. Please add illustrative cases with discussion when algorithm fails.

Fails of algorithm are shown in Fig. 12b and are described on:

• page 8 Data analysis methods, line 290-294: “Setting the search area instead of the exact location of the trace allows for detection similar to the real one, when another point with a large gradient (object boundary, body parts with extreme temperature) located in the close distance from the trace is detected with better score (see Figure 12b).”
• page 10 Results, line 346-347: “In fact, a negative result was often caused by one wrong measurement and/or a small number of all measurements (see Figure 12b cooling).”
• page 13 Discussion, line 404-411: “If there are other regions with a high temperature gradient in the trace neighborhood, then in moments of lower visibility of the trace the temperatures of these elements could be considered as an observation (see Figure 12b heating). As the animal moves, the motionless trace placed on the fur changes its position in relation to other characteristic points and body boundaries. In practice, it turned out that the most common cause of detection errors was the proximity of the object boundary. The incorrect detection was also observed during the analysis of the saliva traces located on the head, where there are numerous points with extreme temperatures.”
• page 14 Conclusions, line 454-458: ”It was also verified that it is not possible to effectively fit the model to data when the saliva trace is not observable in the vast majority of frames (due to the intensive animals’ interaction). In such situations the mutual interaction between animals could be analyzed (e.g. using instance segmentation) and other data analysis methods should be investigated.”

Minors:
2. Fig. 8 quality of images is low - there are some vertical lines visible

Thank you for your careful attention, these lines were unnecessary there. The image has been corrected.

3. Tab.1 \times not * operator

Table 1 presents the results of three-way ANOVA, the operator “*” means the interaction between two (or more) variables. The operator "*" is most often used, less frequently replaced with "x" or "#".

4. Eq.4 brackets size too small

Thank you for your attention. The equation has been corrected.

Round 2

Reviewer 4 Report

ok