Introducing the Segmented HAM10000 (\"Segmented Human Against Machine with 10000 training images\") Dataset (Pre-processed)

Contributors towards pre-processing the existing dataset from the same (dataverse.harvard.edu) platform: Students of College: U.V.Patel College of Engineering, Ganpat University, Gujarat Nishant Darji Kiran Patel Deep Patel Guide: Dr. Yagnesh Vyas During Internship at Bhaskaracharya National Institute for Space Applications and Geo-informatics (BISAG-N) Addressing the challenge of training neural networks for automated diagnosis of pigmented skin lesions, we proudly present the Segmented HAM10000 dataset. Stemming from the original HAM10000, this segmented iteration aims to further enhance the efficacy and diversity of available data for academic machine-learning endeavors. Derived from a diverse array of populations and captured through various modalities, the Segmented HAM10000 dataset boasts a comprehensive collection of 10,015 dermatoscopic images. Each image is meticulously segmented, optimizing it for precise analysis and interpretation by machine learning algorithms. Representing a spectrum of crucial diagnostic categories within pigmented lesions, our dataset encompasses Actinic keratoses and intraepithelial carcinoma / Bowen's disease (akiec), basal cell carcinoma (bcc), benign keratosis-like lesions (solar lentigines / seborrheic keratoses and lichen-planus like keratoses, bkl), dermatofibroma (df), melanoma (mel), melanocytic nevi (nv), and vascular lesions (angiomas, angiokeratomas, pyogenic granulomas, and hemorrhage, vasc). More than 50% of lesions in our dataset are confirmed through histopathology (histo), while the ground truth for the remaining cases is established through follow-up examination (follow_up), expert consensus (consensus), or confirmation by in-vivo confocal microscopy (confocal). Notably, the Segmented HAM10000 dataset includes lesions with multiple images, facilitating tracking via the lesion_id-column within the accompanying metadata file.