Toan Dinh Minh
mtoan65
- Vietnam
- Da Nang University of Science and Technology
- Information Technology Faculty
Statistics
- Member for 1 year, 1 month
- 15 challenge submissions
Activity Overview
ToothFairy2: Multi-Structure Segmentation in CBCT Volumes
Challenge UserThis is the second edition of the ToothFairy challenge organized by the University of Modena and Reggio Emilia with the collaboration of Radboud University Medical Center. The challenge is hosted by grand-challenge and is part of MICCAI2024.
PUMA: Panoptic segmentation of nUclei and tissue in MelanomA
Challenge UserThe PUMA Challenge aims to enhance nuclei and tissue segmentation in melanoma histopathology, addressing the need for better prognostic biomarkers to predict treatment responses. Melanoma, a highly aggressive skin cancer, often requires immune checkpoint inhibition therapy, but only half of patients respond. Prognostic biomarkers like tumor infiltrating lymphocytes (TILs) correlate with better therapy responses and lower recurrence rate, but manual TIL scoring is subjective and inconsistent. Current deep learning methods underperform. The PUMA dataset includes annotated primary and metastatic melanoma regions to improve segmentation techniques. The challenge includes two tracks with tasks focused on tissue and nuclei segmentation, encouraging advanced methods to improve predictive accuracy.
HEad and neCK TumOR Lesion Segmentation, Diagnosis and Prognosis
Challenge UserHECKTOR 2025 is the next iteration of a medical imaging challenge focused on improving automated analysis of head and neck cancer using multimodal PET/CT data. The challenge features three complementary tasks that span the clinical workflow: automatic detection and segmentation of primary tumors and lymph nodes, prediction of recurrence-free survival using imaging and clinical data, and diagnosis of HPV status, which is crucial for treatment decisions. The 2025 edition significantly expands on previous challenges with a larger dataset exceeding, refined evaluation metrics that better assess both detection and segmentation capabilities, and the addition of radiotherapy planning dose maps as an information channel. This challenge aims to advance the development of clinical tools that can aid in treatment planning, outcome prediction, and diagnosis in head and neck cancer patients, ultimately supporting more personalized patient management approaches.
REport Generation in pathology using Pan-Asia Giga-pixel WSIs
Challenge UserThis project focuses on advancing automated pathology report generation using vision-language foundation models. It addresses the limitations of traditional NLP metrics (e.g., BLEU, METEOR, ROUGE) by emphasizing clinically relevant evaluation. The initiative includes standardized datasets, expert comparisons, and medical-domain-specific metrics to assess model performance. It also explores the integration of generated reports into diagnostic workflows with clinical feedback. To support fairness and generalizability, the challenge dataset comprises ~20,500 cases from six medical centers in Korea, Japan, India, Turkey, and Germany, promoting multicultural and multiethnic medical AI development.
ToothFairy3: Multi-Class Segmentation in CBCT Volumes
Challenge UserToothFairy3, part of the ODIN2025 challenge cluster at MICCAI2025, advances CBCT segmentation with an expanded 77-class dataset and a new emphasis on computational efficiency. It introduces two tasks: a runtime-aware multi-structure segmentation and a novel interactive track for Inferior Alveolar Canal (IAC) segmentation using minimal user input. The challenge supports the development of both automated and prompt-based interactive AI tools to enhance clinical workflows in dentistry and maxillofacial surgery.
