While several previous studies have devised methods for segmentation of polyps, most of these methods are not rigorously assessed on multi-center datasets. Variability due to appearance of polyps from one center to another, difference in endoscopic instrument grades, and acquisition quality result in methods with good performance on in-distribution test data, and poor performance on out-of-distribution or underrepresented samples. Unfair models have serious implications and pose a critical challenge to clinical applications. We adapt an implicit bias mitigation method which leverages Bayesian epistemic uncertainties during training to encourage the model to focus on underrepresented sample regions. We demonstrate the potential of this approach to improve generalisability without sacrificing state-of-the-art performance on a challenging multi-center polyp segmentation dataset (PolypGen) with different centers and image modalities.

Prevalence of gastrointestinal (GI) cancer is growing alarmingly every year leading to a substantial increase in the mortality rate. Endoscopic detection is providing crucial diagnostic support, however, subtle lesions in upper and lower GI are quite hard to detect and cause considerable missed detection. In this work, we leverage deep learning to develop a framework to improve the localization of difficult to detect lesions and minimize the missed detection rate. We propose an end to end student-teacher learning setup where class probabilities of a trained teacher model on one class with larger dataset are used to penalize multi-class student network. Our model achieves higher performance in terms of mean average precision (mAP) on both endoscopic disease detection (EDD2020) challenge and Kvasir-SEG datasets. Additionally, we show that using such learning paradigm, our model is generalizable to unseen test set giving higher APs for clinically crucial neoplastic and polyp categories

In this contribution we use an ensemble deep-learning method for combining the prediction of two individual one-stage detectors (i.e., YOLOv4 and Yolact) with the aim to detect artefacts in endoscopic images. This ensemble strategy enabled us to improve the robustness of the individual models without harming their real-time computation capabilities. We demonstrated the effectiveness of our approach by training and testing the two individual models and various ensemble configurations on the "Endoscopic Artifact Detection Challenge" dataset. Extensive experiments show the superiority, in terms of mean average precision, of the ensemble approach over the individual models and previous works in the state of the art.

This work describes the development of a controlled cabin for capturing and analyzing thermal images. The motivation of such a device is to aid in the thermal image acquisition process within a confined space. The thermograms generated provide helpful information for analyzing the residual human limb in subjects with transtibial amputation. Such a study proposes a non-intrusive method to study the thermal activity on the amputee residual limb and seek a correlation to the quality of the socket. The proposed cabin ensures the repeatability of the thermograms acquisition process and provides an isolated workspace, thus improving the quality of the samples. The methodology consists of the design of the mechanical elements and parts of the system on computer-aided design software, the electronic instrumentation, a graphic user interface, and the control algorithm based on a barrier Lyapunov function to solve the trajectory tracking for the camera movements, and numerical simulations to illustrate the functionality and the manufacture of a prototype. The results obtained by implementing the control design on the automated cabin reveal that the thermal image acquisition process is completed following the desired trajectory with a mean squared tracking error of 0.0052. In addition, an example of the thermal images of two subjects and the results processing this class of pictures using the designed interface is shown.

Colorectal cancer cases have been increasing at an alarming rate each year, imposing a healthcare burden worldwide. Multiple efforts have been made to treat this malignancy. However, early screening has been the most promising solution. Optical endoscopy is the primary diagnosis and treatment tool for these malignancies. Even though its success, the endoscopic process represents a challenge due to noisy data, a limited field of view and the presence of multiple artefacts. In this work, we present a comparison between two real-time deep learning frameworks trained to detect artefacts in endoscopic data. Both networks are trained using different data augmentation techniques to analyze their effect when the models are evaluated using data coming from a different distribution. We evaluated these models using the mean average precision (mAP) evaluation metric at a different intersection over union values. Both models outperformed state-of-the-art methods that were evaluated using the same dataset. Also, the use of data augmentation techniques showed an overall improvement in terms of mAP when compared to the case in which no augmentation was applied.