How prajxal/parkinson-s-disease-detection-with-qsvm works
This repository is a Parkinson’s detection research pipeline: MRI images are catalogued, converted to CNN embeddings, pooled by subject, then evaluated with classical classifiers and a quantum-kernel SVM. The main user entry point, src/run_experiments.py, coordinates those stages. The README also describes a voice-data classical baseline and embedding EDA; the checked-in voice data and EDA generator support those capabilities, though their full runtime wiring is not established by the sampled excerpts. Results include metrics and plots. The graph focuses on supported calls and documented capabilities; unsampled subsystem links are omitted.
Main entry points: pca_qs_pipeline.py PCA and cross-validation · run_experiments.py Experiment runner · cnn_feature_extractor.py CNN feature extraction · embeddings_utils.py Subject aggregation
Components 19
Data inputs
- MRI image dataset
- Voice recordings data
parkinsons/parkinsons.data
Image features
- Experiment runner
src/run_experiments.py - Image manifest
src/data_loader.py - Image dataset loader
src/data_loader.py - CNN feature extraction
src/cnn_feature_extractor.py - Slice embeddings
- Subject aggregation
src/embeddings_utils.py - Subject embeddings
Model evaluation
- PCA and cross-validation
src/pca_qs_pipeline.py - Classical classifiers
src/pca_qs_pipeline.py - Quantum kernel
src/quantum_kernel.py - QSVM and Nyström
src/pca_qs_pipeline.py - Evaluation metrics
src/eval_metrics.py
EDA and results
- Experiment results
- EDA notebook generator
generate_eda_notebook.py - Embedding EDA
notebooks/eda.ipynb - Result visualizations
generate_visuals.py
Other components
- Researcher
Connections 25
- Researcher → Experiment runner: starts
- Experiment runner → CNN feature extraction: runs extraction
- Experiment runner → Subject aggregation: aggregates
- Experiment runner → PCA and cross-validation: runs evaluation
- MRI image dataset → Image manifest: scanned for images
- CNN feature extraction → Image manifest: builds if missing
- CNN feature extraction → Image dataset loader: loads images
- Image dataset loader → Image manifest: reads records
- CNN feature extraction → Slice embeddings: writes
- Subject aggregation → Slice embeddings: reads
- Subject aggregation → Image manifest: reads labels
- Subject aggregation → Subject embeddings: writes
- PCA and cross-validation → Subject embeddings: loads features
- PCA and cross-validation → Classical classifiers: trains
- PCA and cross-validation → Quantum kernel: computes kernels
- PCA and cross-validation → QSVM and Nyström: trains
- PCA and cross-validation → Evaluation metrics: computes metrics
- PCA and cross-validation → Experiment results: writes results
- Researcher → EDA notebook generator: requests analysis
- EDA notebook generator → Embedding EDA: generates notebook
- Embedding EDA → Slice embeddings: analyzes embeddings
- Embedding EDA → Subject embeddings: analyzes embeddings
- Researcher → Result visualizations: requests plots
- Result visualizations → Experiment results: reads metrics
- Voice recordings data → Classical classifiers: classical baseline data