{
"cells": [
{
"cell_type": "markdown",
"id": "e7218e6e",
"metadata": {},
"source": [
"# Synthetic Pairing Design\n",
"\n",
"## Objective\n",
"This notebook builds the artificial pairing setup that links the corrected BreaKHis image branch to the unchanged Wisconsin tabular branch. My first draft of the pairing logic was simply label aligned, and the numbers looked clean enough to make me suspicious, so this notebook is also where I make the control logic explicit.\n"
]
},
{
"cell_type": "markdown",
"metadata": {
"codex_research_commentary": true
},
"source": [
"## Notebook Purpose\n",
"\n",
"This notebook constructs the synthetic pairing design used for exploratory multimodal experiments. It pairs Wisconsin tabular records with BreaKHis patient-level image embeddings under controlled strategies.\n"
]
},
{
"cell_type": "markdown",
"metadata": {
"codex_research_commentary": true
},
"source": [
"## Why This Matters\n",
"\n",
"Because true matched multimodal patient data is not available, pairing must be treated as a methodological simulation. This notebook makes the pairing choices transparent and introduces random-pairing controls to avoid overclaiming same-label results.\n"
]
},
{
"cell_type": "markdown",
"metadata": {
"codex_research_commentary": true
},
"source": [
"## Build Pairing Inputs\n",
"\n",
"This setup cell loads the published Wisconsin dataframe and creates patient-level BreaKHis image embeddings. Aggregating image embeddings by patient keeps the image side aligned with the corrected patient-level protocol.\n"
]
},
{
"cell_type": "code",
"execution_count": 1,
"id": "b1309631",
"metadata": {
"execution": {
"iopub.execute_input": "2026-04-19T22:41:28.653735Z",
"iopub.status.busy": "2026-04-19T22:41:28.653556Z",
"iopub.status.idle": "2026-04-19T22:42:23.316597Z",
"shell.execute_reply": "2026-04-19T22:42:23.315811Z"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Project root: /Users/sergeysotskiy/Documents/UNI/year 3/Dissertation/dissertation_project\n",
"Outputs: /Users/sergeysotskiy/Documents/UNI/year 3/Dissertation/dissertation_project/outputs\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Wisconsin rows: 569\n",
"Image embedding rows: 82\n",
"Unique image patients: 82\n"
]
}
],
"source": [
"from __future__ import annotations\n",
"\n",
"import json\n",
"import random\n",
"import sys\n",
"from pathlib import Path\n",
"\n",
"import matplotlib.pyplot as plt\n",
"import numpy as np\n",
"import pandas as pd\n",
"\n",
"SEED = 42\n",
"random.seed(SEED)\n",
"np.random.seed(SEED)\n",
"plt.style.use('seaborn-v0_8-whitegrid')\n",
"\n",
"CWD = Path.cwd().resolve()\n",
"if (CWD / 'src').exists() and (CWD / 'data').exists():\n",
" PROJECT_ROOT = CWD\n",
"elif (CWD.parent / 'src').exists() and (CWD.parent / 'data').exists():\n",
" PROJECT_ROOT = CWD.parent\n",
"elif (CWD.parent.parent / 'src').exists() and (CWD.parent.parent / 'data').exists():\n",
" PROJECT_ROOT = CWD.parent.parent\n",
"else:\n",
" raise RuntimeError(f'Could not resolve dissertation_project root from {CWD}')\n",
"\n",
"REPO_ROOT = PROJECT_ROOT.parent\n",
"OUTPUTS = PROJECT_ROOT / 'outputs'\n",
"FIGURES = OUTPUTS / 'figures'\n",
"METRICS = OUTPUTS / 'metrics'\n",
"REPORTS = OUTPUTS / 'reports'\n",
"MODELS = PROJECT_ROOT / 'models'\n",
"DATA_ROOT = PROJECT_ROOT / 'data' / 'dataset_cancer_v1' / 'dataset_cancer_v1'\n",
"WISCONSIN_ROOT = PROJECT_ROOT / 'notebook_Wisconsin'\n",
"\n",
"for path in [FIGURES, METRICS, REPORTS]:\n",
" path.mkdir(parents=True, exist_ok=True)\n",
"\n",
"if str(PROJECT_ROOT) not in sys.path:\n",
" sys.path.append(str(PROJECT_ROOT))\n",
"\n",
"print('Project root:', PROJECT_ROOT)\n",
"print('Outputs:', OUTPUTS)\n",
"\n",
"from IPython.display import display\n",
"\n",
"from src.breakhis import build_binary_records, patient_level_split\n",
"from src.fusion import aggregate_image_embeddings, extract_image_embeddings, load_wisconsin_dataframe\n",
"from src.pairing import add_consistency_flag, build_random_pairs, build_same_label_pairs\n",
"\n",
"wisconsin = load_wisconsin_dataframe(WISCONSIN_ROOT / 'brca.csv')\n",
"df = build_binary_records(DATA_ROOT)\n",
"split = patient_level_split(df, random_state=SEED)\n",
"image_source = pd.concat([split.train, split.val, split.test], ignore_index=True)\n",
"image_embeddings = aggregate_image_embeddings(\n",
" extract_image_embeddings(image_source, MODELS / 'breakhis_resnet18_patient_level_clean.pth', batch_size=32)\n",
")\n",
"image_embeddings.to_csv(REPORTS / 'breakhis_patient_level_embeddings.csv', index=False)\n",
"print('Wisconsin rows:', len(wisconsin))\n",
"print('Image embedding rows:', len(image_embeddings))\n",
"print('Unique image patients:', image_embeddings['patient_id'].nunique())\n"
]
},
{
"cell_type": "markdown",
"metadata": {
"codex_research_commentary": true
},
"source": [
"## Compare Same-Label And Random Pairing\n",
"\n",
"This cell creates one same-label pairing and one random pairing. The `pair_label_match` flag quantifies how much artificial label agreement each strategy introduces.\n"
]
},
{
"cell_type": "code",
"execution_count": 2,
"id": "31a70c54",
"metadata": {
"execution": {
"iopub.execute_input": "2026-04-19T22:42:23.328384Z",
"iopub.status.busy": "2026-04-19T22:42:23.328210Z",
"iopub.status.idle": "2026-04-19T22:42:23.351465Z",
"shell.execute_reply": "2026-04-19T22:42:23.351008Z"
}
},
"outputs": [
{
"data": {
"text/html": [
"\n",
"\n",
"
\n",
" \n",
" \n",
" pairing_type \n",
" match_rate \n",
" \n",
" \n",
" \n",
" \n",
" 0 \n",
" same_label \n",
" 1.000000 \n",
" \n",
" \n",
" 1 \n",
" random \n",
" 0.428822 \n",
" \n",
" \n",
"
\n",
"
"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"fig, axes = plt.subplots(1, 2, figsize=(10, 4))\n",
"axes[0].bar(consistency_summary['pairing_type'], consistency_summary['match_rate'], color=['#31a354', '#de2d26'])\n",
"axes[0].set_ylim(0, 1.05)\n",
"axes[0].set_title('Pair label agreement rate')\n",
"axes[0].set_ylabel('Match rate')\n",
"\n",
"same_counts = same_label_pairs['img_label'].value_counts().rename_axis('label').reset_index(name='count')\n",
"random_counts = random_pairs['img_label'].value_counts().rename_axis('label').reset_index(name='count')\n",
"axes[1].bar(same_counts['label'], same_counts['count'], alpha=0.7, label='same_label')\n",
"axes[1].bar(random_counts['label'], random_counts['count'], alpha=0.7, label='random')\n",
"axes[1].set_title('Image-label distribution after pairing')\n",
"axes[1].legend()\n",
"fig.tight_layout()\n",
"fig.savefig(FIGURES / 'synthetic_pairing_design_summary.png', dpi=300)\n",
"plt.show()\n"
]
},
{
"cell_type": "markdown",
"metadata": {
"codex_research_commentary": true
},
"source": [
"## Repeated-Seed Pairing Preview\n",
"\n",
"This cell repeats the pairing process across seeds. The goal is to show whether match rates are stable or seed-dependent before running full fusion experiments.\n"
]
},
{
"cell_type": "code",
"execution_count": 4,
"id": "043fe759",
"metadata": {
"execution": {
"iopub.execute_input": "2026-04-19T22:42:23.604277Z",
"iopub.status.busy": "2026-04-19T22:42:23.604179Z",
"iopub.status.idle": "2026-04-19T22:42:23.793170Z",
"shell.execute_reply": "2026-04-19T22:42:23.792681Z"
}
},
"outputs": [
{
"data": {
"text/html": [
"\n",
"\n",
"
\n",
" \n",
" \n",
" seed \n",
" pairing_type \n",
" match_rate \n",
" \n",
" \n",
" \n",
" \n",
" 0 \n",
" 11 \n",
" same_label \n",
" 1.000000 \n",
" \n",
" \n",
" 1 \n",
" 11 \n",
" random \n",
" 0.428822 \n",
" \n",
" \n",
" 2 \n",
" 21 \n",
" same_label \n",
" 1.000000 \n",
" \n",
" \n",
" 3 \n",
" 21 \n",
" random \n",
" 0.439367 \n",
" \n",
" \n",
" 4 \n",
" 31 \n",
" same_label \n",
" 1.000000 \n",
" \n",
" \n",
" 5 \n",
" 31 \n",
" random \n",
" 0.483304 \n",
" \n",
" \n",
" 6 \n",
" 41 \n",
" same_label \n",
" 1.000000 \n",
" \n",
" \n",
" 7 \n",
" 41 \n",
" random \n",
" 0.441125 \n",
" \n",
" \n",
" 8 \n",
" 51 \n",
" same_label \n",
" 1.000000 \n",
" \n",
" \n",
" 9 \n",
" 51 \n",
" random \n",
" 0.414763 \n",
" \n",
" \n",
"
\n",
"