"Towards Rapid Mycetoma Species Diagnosis: A Deep Learning Approach for Stain-Invariant Classification on H&E Images from Senegal," introduces an innovative automated system for identifying mycetoma species using histopathological images of patients with black skin in Senegal. This research fills a critical void in the field, as no dataset had previously been available for mycetoma diagnostics in this context. Moreover, the challenge of stain variability in existing histopathological methods has further complicated accurate diagnosis, particularly in low-resource settings. Given that mycetoma is a neglected tropical disease requiring precise type and species identification for effective treatment, we meticulously curated, annotated, and labeled a unique dataset of Hematoxylin and Eosin (H&E)-stained images from various hospitals across Senegal. By employing advanced stain normalization techniques such as Macenko, Vahadane, and Reinhard, in conjunction with a deep learning model utilizing the MONAI framework and DenseNet121 architecture, our system achieves exceptional classification accuracy while significantly enhancing diagnostic reliability. This work represents a vital advancement in mycetoma diagnosis, aiming to improve the speed and accuracy of medical interventions in endemic regions.
Presenting my paper at MICCAI 2024 is a significant milestone for me, both personally and professionally. It provides an invaluable opportunity to share my research with a global audience of experts and practitioners in the field of medical imaging. Receiving the African Travel Grant enhances this experience by facilitating my attendance and highlighting the importance of supporting African researchers in the global scientific community. This opportunity not only enables me to engage in important discussions but also facilitates connections with fellow researchers, allowing for meaningful feedback on my work, and potential collaborations that could enhance its impact.
I am particularly interested in learning more about advancements in deep learning techniques applied to medical imaging, especially those focused on improving diagnostic accuracy in resource-limited settings. Additionally, I am eager to explore discussions surrounding algorithmic fairness and human-centered AI in medical imaging, as these topics are crucial for ensuring equitable healthcare solutions. I also hope to gain insights into the integration of imaging and non-imaging data for personalized medicine, which could enhance the relevance of my own research.
I am looking forward to the vibrant atmosphere of MICCAI 2024 in Marrakesh, where I hope to engage with fellow researchers and industry leaders who share a passion for medical image computing. Attending various presentations and workshops will be an exciting opportunity to learn about cutting-edge research and innovations in the field. I am particularly excited about networking with other attendees, exchanging ideas, and discussing potential collaborations. Additionally, experiencing the rich culture and history of Marrakesh will add to the overall experience of the conference.