How yashjoshi89/8-sem-project works
This project loads and normalizes DICOM CT images, trains a CNN for PE classification, and evaluates predictions. Its interactive path lets a user upload a DICOM image, runs classification and a separate image-based location estimate, then displays both results. The graph also includes checkpoint persistence. The README describes prediction and training entry points not present in the supplied tree; those are not represented as separate modules. Connections reflect the shown source and README, not uninspected behavior.
Main entry points: model.py PE CNN · main.py Training Runner · evaluate.py Evaluator · gui.py Analysis GUI
Components 8
DICOM Data
- DICOM Loader
data_loader.py
Model and Evaluation
- Training Runner
main.py - PE CNN
model.py - Model Checkpoint
- Evaluator
evaluate.py
Interactive Analysis
- Analysis GUI
gui.py - Location Detector
location_detector.py
Other components
- User
Connections 10
- Training Runner → DICOM Loader: loads training data
main.py - Training Runner → PE CNN: trains and saves
main.py - PE CNN → Model Checkpoint: writes checkpoint
model.py - PE CNN → Model Checkpoint: restores checkpoint
model.py - Training Runner → Evaluator: runs evaluation
main.py - Evaluator → DICOM Loader: loads test data
evaluate.py - Evaluator → PE CNN: loads and predicts
evaluate.py - User → Analysis GUI: uploads DICOM
gui.py - Analysis GUI → PE CNN: classifies image
gui.py - Analysis GUI → Location Detector: detects location
gui.py