Week3ProjectPublic

BWSI Medlytics 2022 - Challenge Project for Week 3 (Image Processing and Advanced Analytics)

• Contains a notebook with the challenge details explained fully. In this challenge students classify segments of mammogram images as 0 (negative for abnormality) or 1 (positive for abnormality) or 0, 1, 2, 3, 4 in the multiclass problem (negative and varying types of abnormalities).

• Dataset used in this challenge: Mammogram images from the Digital Database for Screening Mammography (DDSM) and the Curated Breast Imaging Subset of DDSM (CBIS-DDSM). Both datasets are publicly available:

  • DDSM - http://marathon.csee.usf.edu/Mammography/Database.html
  • CBIS-DDSM - https://wiki.cancerimagingarchive.net/display/Public/CBIS-DDSM

• Jpegs and labels were extracted from a tfrecords dataset curated by Eric Scuccimarra (https://www.kaggle.com/skooch)

• Created by: Thomas Possidente.