Clinically Significant Prostate Cancer Detection in bpMRI


Related Publications:

Source Code: https://github.com/DIAGNijmegen/prostateMR_3D-CAD-csPCa/

Contact: anindya.shaha@radboudumc.nl, matin.hosseinzadeh@radboudumc.nl


Description: This algorithm is a deep learning-based detection/diagnosis model (deep attentive 3D U-Net), which was trained, validated and tested using 2732 prostate biparametric MRI (bpMRI) scans paired with PI-RADS v2 annotations. Its input requires 3 axial/transverse MRI volumes in .mha format:

  1. T2-weighted imaging (T2W)
  2. High b-value diffusion-weighted imaging (HBV)
  3. Apparent diffusion coefficient map (ADC)

In return, the model predicts a single heatmap for the likelihood of clinically significant prostate cancer.

Important Notes:
● This algorithm is intended for research purposes only. Further advancements are required before direct clinical usage.
● This algorithm was trained and evaluated exclusively on prostate bpMRI scans derived from Siemens Healthineers (Skyra/Prisma) 3T MRI scanners with surface coils. It does not account for vendor-neutral properties or domain adaptation, and in turn, is not compatible for scans derived using any other MRI scanner or those using endorectal coils.
● This algorithm was trained on biopsy-naïve patients that had not undergone any prior treatments (e.g. radiotherapy, transurethral resection of the prostate (TURP), cryoablation, etc.). Hence it is susceptible to faulty predictions and unintended behaviour, when presented with cases that do not belong to this distribution.
● While the input images (T2W, HBV, ADC) can be of different spatial resolutions, the algorithm assumes that they are co-registered or aligned reasonably well, and that the prostate gland is localized within a volume of 336 cm³ from the center coordinate.