Deep transfer learning can be used for the detection of hip joints in pelvis radiographs and the classification of their hip dysplasia status

Artificial intelligence in veterinary diagnostic imaging: Perspectives and limitations

The field of veterinary diagnostic imaging is undergoing significant transformation with the integration of artificial intelligence (AI) tools. This manuscript provides an overview of the current state and future prospects of AI in veterinary diagnostic imaging. The manuscript delves into various applications of AI across different imaging modalities, such as radiology, ultrasound, computed tomography, and magnetic resonance imaging. Examples of AI applications in each modality are provided, ranging from orthopaedics to internal medicine, cardiology, and more. Notable studies are discussed, demonstrating AI's potential for improved accuracy in detecting and classifying various abnormalities. The ethical considerations of using AI in veterinary diagnostics are also explored, highlighting the need for transparent AI development, accurate training data, awareness of the limitations of AI models, and the importance of maintaining human expertise in the decision-making process. The manuscript underscores the significance of AI as a decision support tool rather than a replacement for human judgement. In conclusion, this comprehensive manuscript offers an assessment of the current landscape and future potential of AI in veterinary diagnostic imaging. It provides insights into the benefits and challenges of integrating AI into clinical practice while emphasizing the critical role of ethics and human expertise in ensuring the wellbeing of veterinary patients.

Deep learning models for interpretation of point of care ultrasound in military working dogs

Introduction

Military working dogs (MWDs) are essential for military operations in a wide range of missions. With this pivotal role, MWDs can become casualties requiring specialized veterinary care that may not always be available far forward on the battlefield. Some injuries such as pneumothorax, hemothorax, or abdominal hemorrhage can be diagnosed using point of care ultrasound (POCUS) such as the Global FAST® exam. This presents a unique opportunity for artificial intelligence (AI) to aid in the interpretation of ultrasound images. In this article, deep learning classification neural networks were developed for POCUS assessment in MWDs.

Methods

Images were collected in five MWDs under general anesthesia or deep sedation for all scan points in the Global FAST® exam. For representative injuries, a cadaver model was used from which positive and negative injury images were captured. A total of 327 ultrasound clips were captured and split across scan points for training three different AI network architectures: MobileNetV2, DarkNet-19, and ShrapML. Gradient class activation mapping (GradCAM) overlays were generated for representative images to better explain AI predictions.

Results

Performance of AI models reached over 82% accuracy for all scan points. The model with the highest performance was trained with the MobileNetV2 network for the cystocolic scan point achieving 99.8% accuracy. Across all trained networks the diaphragmatic hepatorenal scan point had the best overall performance. However, GradCAM overlays showed that the models with highest accuracy, like MobileNetV2, were not always identifying relevant features. Conversely, the GradCAM heatmaps for ShrapML show general agreement with regions most indicative of fluid accumulation.

Discussion

Overall, the AI models developed can automate POCUS predictions in MWDs. Preliminarily, ShrapML had the strongest performance and prediction rate paired with accurately tracking fluid accumulation sites, making it the most suitable option for eventual real-time deployment with ultrasound systems. Further integration of this technology with imaging technologies will expand use of POCUS-based triage of MWDs.

Evaluation of a Novel Veterinary Dental Radiography Artificial Intelligence Software Program

There is a growing trend of artificial intelligence (AI) applications in veterinary medicine, with the potential to assist veterinarians in clinical decisions. A commercially available, AI-based software program (AISP) for detecting common radiographic dental pathologies in dogs and cats was assessed for agreement with two human evaluators. Furcation bone loss, periapical lucency, resorptive lesion, retained tooth root, attachment (alveolar bone) loss and tooth fracture were assessed. The AISP does not attempt to diagnose or provide treatment recommendations, nor has it been trained to identify other types of radiographic pathology. Inter-rater reliability for detecting pathologies was measured by absolute percent agreement and Gwet's agreement coefficient. There was good to excellent inter-rater reliability among all raters, suggesting the AISP performs similarly at detecting the specified pathologies compared to human evaluators. Sensitivity and specificity for the AISP were assessed using human evaluators as the reference standard. The results revealed a trend of low sensitivity and high specificity, suggesting the AISP may produce a high rate of false negatives and may not be a good tool for initial screening. However, the low rate of false positives produced by the AISP suggests it may be beneficial as a "second set of eyes" because if it detects the specific pathology, there is a high likelihood that the pathology is present. With an understanding of the AISP, as an aid and not a substitute for veterinarians, the technology may increase dental radiography utilization and diagnostic potential.

Femoral Neck Thickness Index as an Indicator of Proximal Femur Bone Modeling

The alteration in the shape of the femoral neck is an important radiographic sign for scoring canine hip dysplasia (CHD). Previous studies have reported that the femoral neck thickness (FNT) is greater in dogs with hip joint dysplasia, becoming progressively thicker with disease severity. The main objective of this work was to describe a femoral neck thickness index (FNTi) to quantify FNT and to study its association with the degree of CHD using the Fédération Cynologique Internationale (FCI) scheme. A total of 53 dogs (106 hips) were randomly selected for this study. Two examiners performed FNTi estimation to study intra- and inter-examiner reliability and agreement. The paired t-test, the Bland-Altman plots, and the intraclass correlation coefficient showed excellent agreement and reliability between the measurements of the two examiners and the examiners' sessions. All joints were scored in five categories by an experienced examiner according to FCI criteria. The results from examiner 1 were compared between FCI categories. Hips that were assigned an FCI grade of A (n = 19), B (n = 23), C (n = 24), D (n = 24), and E (n = 16) had a mean ± standard deviation FNTi of 0.809 ± 0.024, 0.835 ± 0.044, 0.868 ± 0.022, 0.903 ± 0.033, and 0.923 ± 0.068, respectively (ANOVA, p < 0.05). Therefore, these results show that FNTi is a parameter capable of evaluating proximal femur bone modeling and that it has the potential to enrich conventional CHD scoring criteria if incorporated into a computer-aided diagnosis capable of detecting CHD.

Artificial Intelligence in Veterinary Imaging: An Overview

Artificial intelligence and machine learning have been increasingly used in the medical imaging field in the past few years. The evaluation of medical images is very subjective and complex, and therefore the application of artificial intelligence and deep learning methods to automatize the analysis process would be very beneficial. A lot of researchers have been applying these methods to image analysis diagnosis, developing software capable of assisting veterinary doctors or radiologists in their daily practice. This article details the main methodologies used to develop software applications on machine learning and how veterinarians with an interest in this field can benefit from such methodologies. The main goal of this study is to offer veterinary professionals a simple guide to enable them to understand the basics of artificial intelligence and machine learning and the concepts such as deep learning, convolutional neural networks, transfer learning, and the performance evaluation method. The language is adapted for medical technicians, and the work already published in this field is reviewed for application in the imaging diagnosis of different animal body systems: musculoskeletal, thoracic, nervous, and abdominal.

The Role of Femoral Head Size and Femoral Head Coverage in Dogs with and without Hip Dysplasia

The subject of hip dysplasia in dogs is still current and preoccupies both animal owners and veterinarians. Major factors affecting the development of the disorder are hip laxity and incongruent joints. Many studies on etiology, pathogenesis, and early diagnosis have been performed to reduce prevalence and select healthy dogs for breeding. The purpose of the present study was to investigate a possible relationship between dysplasia and femoral head area (FHA), femoral coverage by the acetabulum (CFH) and cranio-caudal distance of the dorsal acetabular rim (CrCdAR). Radiographs of a total of 264 skeletally mature dogs with similar physical characteristics (German wirehaired pointers (GWP), German shepherd dogs (GSD) and Labrador retrievers (LAB)) presented for routine hip dysplasia screening were recruited for the study. FHA, CFH and CrCdAR were measured and related to dysplasia status. Evaluations of FHA (p = 0.011), CFH (p < 0.001) and CrCdAR length (p = 0.003) measurements revealed significant interactions between breed, sex and FCI scores, so they had to be assessed separately. The results revealed that FHA tends to decrease as the hip dysplasia score worsens. There was no significant relationship between FHA and dysplasia assessment. FHA is breed-specific and is larger in normal and near-normal male (p = 0.001, p = 0.020) and female (p = 0.001, p = 0.013) GWP compared to GSD, respectively. FHA is greater in normal male GWP (p = 0.011) and GSD (p = 0.040) compared to females. There was a significant and strong positive correlation between FHA and CrCdAR in all breeds and sexes. Additionally, FCI scoring had a medium (GWP, GSD) to strong (LAB) negative correlation with CFH.

Artificial intelligence in veterinary diagnostic imaging: A literature review

Artificial intelligence in veterinary medicine is an emerging field. Machine learning, a subfield of artificial intelligence, allows computer programs to analyze large imaging datasets and learn to perform tasks relevant to veterinary diagnostic imaging. This review summarizes the small, yet growing body of artificial intelligence literature in veterinary imaging, provides necessary background to understand these papers, and provides author commentary on the state of the field. To date, less than 40 peer-reviewed publications have utilized machine learning to perform imaging-associated tasks across multiple anatomic regions in veterinary clinical and biomedical research. Major challenges in this field include collection and cleaning of sufficient image data, selection of high-quality ground truth labels, formation of relationships between veterinary and machine learning professionals, and closure of the gap between academic uses of artificial intelligence and currently available commercial products. Further development of artificial intelligence has the potential to help meet the growing need for radiological services through applications in workflow, quality control, and image interpretation for both general practitioners and radiologists.

Acetabular Coverage Area Occupied by the Femoral Head as an Indicator of Hip Congruency

Accurate radiographic screening evaluation is essential in the genetic control of canine HD, however, the qualitative assessment of hip congruency introduces some subjectivity, leading to excessive variability in scoring. The main objective of this work was to validate a method-Hip Congruency Index (HCI)-capable of objectively measuring the relationship between the acetabulum and the femoral head and associating it with the level of congruency proposed by the Fédération Cynologique Internationale (FCI), with the aim of incorporating it into a computer vision model that classifies HD autonomously. A total of 200 dogs (400 hips) were randomly selected for the study. All radiographs were scored in five categories by an experienced examiner according to FCI criteria. Two examiners performed HCI measurements on 25 hip radiographs to study intra- and inter-examiner reliability and agreement. Additionally, each examiner measured HCI on their half of the study sample (100 dogs), and the results were compared between FCI categories. The paired t-test and the intraclass correlation coefficient (ICC) showed no evidence of a systematic bias, and there was excellent reliability between the measurements of the two examiners and examiners’ sessions. Hips that were assigned an FCI grade of A (n = 120), B (n = 157), C (n = 68), D (n = 38) and E (n = 17) had a mean HCI of 0.739 ± 0.044, 0.666 ± 0.052, 0.605 ± 0.055, 0.494 ± 0.070 and 0.374 ± 0.122, respectively (ANOVA, p < 0.01). Therefore, these results show that HCI is a parameter capable of estimating hip congruency and has the potential to enrich conventional HD scoring criteria if incorporated into an artificial intelligence algorithm competent in diagnosing HD.

Deep learning in veterinary medicine, an approach based on CNN to detect pulmonary abnormalities from lateral thoracic radiographs in cats

Thoracic radiograph (TR) is a complementary exam widely used in small animal medicine which requires a sharp analysis to take full advantage of Radiographic Pulmonary Pattern (RPP). Although promising advances have been made in deep learning for veterinary imaging, the development of a Convolutional Neural Networks (CNN) to detect specifically RPP from feline TR images has not been investigated. Here, a CNN based on ResNet50V2 and pre-trained on ImageNet is first fine-tuned on human Chest X-rays and then fine-tuned again on 500 annotated TR images from the veterinary campus of VetAgro Sup (Lyon, France). The impact of manual segmentation of TR’s intrathoracic area and enhancing contrast method on the CNN’s performances has been compared. To improve classification performances, 200 networks were trained on random shuffles of training set and validation set. A voting approach over these 200 networks trained on segmented TR images produced the best classification performances and achieved mean Accuracy, F1-Score, Specificity, Positive Predictive Value and Sensitivity of 82%, 85%, 75%, 81% and 88% respectively on the test set. Finally, the classification schemes were discussed in the light of an ensemble method of class activation maps and confirmed that the proposed approach is helpful for veterinarians.

A Machine Learning-Based Approach for Classification of Focal Splenic Lesions Based on Their CT Features

The aim of the study was to describe the CT features of focal splenic lesions (FSLs) in dogs in order to predict lesion histotype. Dogs that underwent a CT scan and had a FSL diagnosis by cytology or histopathology were retrospectively included in the study. For the statistical analysis the cases were divided into four groups, based on the results of cytopatholoy or hystopathology, namely: nodular hyperplasia (NH), other benign lesions (OBLs), sarcoma (SA), round cell tumour (RCT). Several qualitative and quantitative CT features were described for each case. The relationship occurring between each individual CT feature and the histopathological groups was explred by means of c chi-square test for the count data and by means of Kruskal-Wallis or ANOVA for the continuous data. Furthermore, the main features of each group were described using factorial discriminant analysis, and a decision tree for lesion classification was then developed. Sarcomas were characterised by large dimensions, a cystic appearance and an overall low post contrast-enhancement. NH and OBLs were characterised by small dimensions, a solid appearance and a high post-contrast enhancement. OBLs showed higher post-contrast values than NH. Lastly, RCTs did not exhibit any distinctive CT features. The proposed decision tree had a high accuracy for the classification of SA (0.89) and a moderate accuracy for the classification of OBLs and NH (0.79), whereas it was unable to classify RCTs. The results of the factorial analysis and the proposed decision tree could help the clinician in classifying FSLs based on their CT features. A definitive FSL diagnosis can only be obtained by microscopic examination of the spleen.

Employing deep convolutional neural networks for segmenting the medial retropharyngeal lymph nodes in CT studies of dogs

While still in its infancy, the application of deep convolutional neural networks in veterinary diagnostic imaging is a rapidly growing field. The preferred deep learning architecture to be employed is convolutional neural networks, as these provide the structure preferably used for the analysis of medical images. With this retrospective exploratory study, the applicability of such networks for the task of delineating certain organs with respect to their surrounding tissues was tested. More precisely, a deep convolutional neural network was trained to segment medial retropharyngeal lymph nodes in a study dataset consisting of CT scans of canine heads. With a limited dataset of 40 patients, the network in conjunction with image augmentation techniques achieved an intersection-overunion of overall fair performance (median 39%, 25 percentiles at 22%, 75 percentiles at 51%). The results indicate that these architectures can indeed be trained to segment anatomic structures in anatomically complicated and breed-related variating areas such as the head, possibly even using just small training sets. As these conditions are quite common in veterinary medical imaging, all routines were published as an open-source Python package with the hope of simplifying future research projects in the community.