All projects
CompletedAI / ML · Research

NeuroTrace

Explainable deep learning for brain-tumor segmentation

NeuroTrace explores brain-tumor segmentation on the BraTS 2023 glioma dataset. The principal architecture is a 3D Attention U-Net with GNN-driven attention at the bottleneck, supported by qualitative explainability and uncertainty work.

Project field / neurotrace

Context

The problem space.

Brain-tumor segmentation is spatially complex and clinically sensitive. The work focuses on a volumetric segmentation pipeline while also making model attention and uncertainty more inspectable.

Confirmed work

What the evidence supports.

  • A 3D Attention U-Net forms the principal segmentation architecture.
  • The bottleneck uses GNN-based, GNN-driven attention rather than Transformer attention.
  • Inputs use patch cropping and z-score normalization on the BraTS 2023 glioma dataset.
  • Training used combined Cross-Entropy and Dice loss with Adam at a learning rate around 1e-4.
  • Grad-CAM and Monte Carlo Dropout support qualitative explainability and uncertainty exploration.
3D Deep LearningMedical Image SegmentationGraph Neural NetworksGrad-CAMMonte Carlo DropoutUncertainty Estimation

System flow

  1. 01

    BraTS 2023 volumes

  2. 02

    Patch cropping

  3. 03

    Z-score normalization

  4. 04

    3D Attention U-Net

  5. 05

    GNN-driven bottleneck attention

  6. 06

    Segmentation output

  7. 07

    Explainability & uncertainty views

Evidence note

Known training context includes patches around 100 × 100 × 60, an RTX 3060 Laptop GPU and approximately 25–50 epochs.

Result limitation

Qualitative segmentation material exists. No quantitative performance metrics are claimed here, and the uncertainty and explainability modules were not evaluated to the same extent as the main segmentation implementation.

Related work

Keep exploring.

All projects
SSProject / 01

AI / ML · Full Stack · Cloud & DevOps

SmartSkin

Skincare Compatibility System

Completed

A skincare compatibility system that evaluates product ingredients against skin type and supports more personal skincare routines.

  • Python
  • Machine Learning
  • React
  • REST APIs
Case notes
View project
PNProject / 02

AI / ML

Personalized News Recommendation System

Semantic news discovery shaped around the reader

Completed

A news recommendation pipeline using text embeddings, topic clustering, preferences and semantic similarity.

  • Python
  • NLP
  • Sentence Transformers
  • Web Scraping
Case notes
View project
TSProject / 03

AI / ML · Full Stack · Cloud & DevOps

TripShield

Autonomous Travel-Disruption Concierge

Ongoing

An ongoing proposal for detecting travel disruption and coordinating bounded, whole-trip recovery rather than only sending alerts.

  • React Native
  • Node.js
  • Express
  • LangGraph
Case notes
View project