IPAB Workshop-02/02/2023

Zhaoxing Deng

Title: Data Collection and Motion Mapping for Skill Learning in Bronchoscopy

Abstract: Bronchoscopy is a common procedure to diagnose and treat lung problems. Due to the complexity in assessment tools and variations in training standards, determining of competency of trainees in performing bronchoscopic procedures can be challenging. Some of the practical skills rely on the experience of expert operators and can be difficult to teach. In this project, we aim to record procedure video data in routine bronchoscopy cases in clinical settings, together with kinematic data from hand and wrist movements of operators, to build a data-driven model which can recognize surgical gestures and assess technical skills in bronchoscopy. This model may help to improve the effectiveness and efficiency in bronchoscopy training and provide guidance towards autonomous robotic bronchoscopy. In this presentation, I will present my recent progress in data collection and motion mapping.

Jack Tsangou

Title: "Pathos: A culturally-diverse affective video dataset"

Abstract: Imbalanced datasets may have worse performance on minority classes or samples with minority features. This effect has been explored on aspects such as race and gender discrepancy. However, cultural significance has not been studied yet, especially within the context of emotion recognition. Therefore, this presentation explores the impact of culture-oriented variance on this field. Current affective datasets have little focus on diversity, thus leading to inaccurate results when involving people who belong in minority groups. In order to mitigate this issue, the present work introduces a publicly available and culturally diverse video dataset.

Zhaoxing Deng

Title: Data Collection and Motion Mapping for Skill Learning in Bronchoscopy

Abstract: Bronchoscopy is a common procedure to diagnose and treat lung problems. Due to the complexity in assessment tools and variations in training standards, determining of competency of trainees in performing bronchoscopic procedures can be challenging. Some of the practical skills rely on the experience of expert operators and can be difficult to teach. In this project, we aim to record procedure video data in routine bronchoscopy cases in clinical settings, together with kinematic data from hand and wrist movements of operators, to build a data-driven model which can recognize surgical gestures and assess technical skills in bronchoscopy. This model may help to improve the effectiveness and efficiency in bronchoscopy training and provide guidance towards autonomous robotic bronchoscopy. In this presentation, I will present my recent progress in data collection and motion mapping.

Jack Tsangou

Title: "Pathos: A culturally-diverse affective video dataset"

Abstract: Imbalanced datasets may have worse performance on minority classes or samples with minority features. This effect has been explored on aspects such as race and gender discrepancy. However, cultural significance has not been studied yet, especially within the context of emotion recognition. Therefore, this presentation explores the impact of culture-oriented variance on this field. Current affective datasets have little focus on diversity, thus leading to inaccurate results when involving people who belong in minority groups. In order to mitigate this issue, the present work introduces a publicly available and culturally diverse video dataset.