--- a
+++ b/notebooks/bland_altman.py
@@ -0,0 +1,147 @@
+# %%
+from pathlib import Path
+
+import matplotlib.pyplot as plt
+import matplotlib.ticker as ticker
+import matplotlib.transforms as transforms
+import numpy as np
+import pandas as pd
+from pyCompare import blandAltman
+
+test_path = "../results/hold-out-test.csv"
+external_all_path = "../results/external-data.csv"
+prospective_all_path = "../results/prospective-data.csv"
+
+figureSize = (4, 4)
+
+
+def dice_plot(
+    x,
+    y,
+    title,
+    figureSize=figureSize,
+    x_label="Reference htTKV (mL)",
+    y_label="Dice Similarity Coefficient",
+):
+
+    dpi = 72
+    fig, ax = plt.subplots(figsize=figureSize, dpi=dpi)
+    draw = True
+    meanColour = "#6495ED"
+    loaColour = "coral"
+    pointColour = "#6495ED"
+
+    ax.scatter(x, y, alpha=0.5, c=pointColour)
+
+    mean = np.mean(y)
+    sd = np.std(y)
+    sd95 = 1.96 * np.std(y)
+    trans = transforms.blended_transform_factory(ax.transAxes, ax.transData)
+
+    offset = 0.08
+    ax.text(
+        0.97,
+        mean - 2 * offset,
+        f"Mean {mean:.2f}",
+        ha="right",
+        va="bottom",
+        transform=trans,
+    )
+    ax.text(
+        0.97,
+        mean - 3 * offset,
+        f"SD {sd:.2f}",
+        ha="right",
+        va="bottom",
+        transform=trans,
+    )
+    ax.text(
+        0.97,
+        mean - 4 * offset,
+        f"±1.96 SD in coral",
+        ha="right",
+        va="bottom",
+        transform=trans,
+    )
+
+    ax.axhspan(mean - sd95, mean + sd95, facecolor=loaColour, alpha=0.2)
+
+    # Hide the right and top spines
+    ax.spines["right"].set_visible(False)
+    ax.spines["top"].set_visible(False)
+
+    ax.set_xlabel(x_label)
+    ax.set_ylabel(y_label)
+    ax.set_title(title, pad=10)
+    ax.set_ylim([0, 1.05])
+    ax.axhline(np.mean(y), color=meanColour, linestyle="--")
+
+    ax.set_yticks(np.arange(0, 1.1, 0.1))
+
+
+# %%
+# external_ds = pd.read_csv(external_path)
+# prospective_ds = pd.read_csv(prospective_path)
+prospective_ds = pd.read_csv(prospective_all_path)
+prospective_ds.TKV_Pred /= 1000  # corrects units to mL
+prospective_ds.TKV_GT /= 1000  # corrects units to mL
+
+test_ds = pd.read_csv(test_path)
+test_ds.TKV_GT /= 1000  # corrects units to mL
+test_ds.TKV_Pred /= 1000  # corrects units to mL
+
+
+external_ds = pd.read_csv(external_all_path)
+# prospective_ds = pd.read_csv(prospective_all_path)
+
+
+# %%
+
+blandAltman(
+    prospective_ds.TKV_GT,
+    prospective_ds.TKV_Pred,
+    percentage=True,
+    title="BA Plot - Prospective dataset",
+    figureSize=figureSize,
+)
+# %%
+blandAltman(
+    external_ds.TKV_GT,
+    external_ds.TKV_Pred,
+    percentage=True,
+    title="BA Plot - External dataset",
+    figureSize=figureSize,
+)
+# %%
+blandAltman(
+    test_ds.TKV_GT,
+    test_ds.TKV_Pred,
+    percentage=True,
+    title="BA Plot - Hold-out-test dataset",
+    figureSize=figureSize,
+)
+
+
+# %%
+dice_plot(
+    prospective_ds.TKV_GT,
+    prospective_ds.patient_dice,
+    title="Dice by TKV - Prospective dataset",
+    figureSize=figureSize,
+)
+
+
+dice_plot(
+    external_ds.TKV_GT,
+    external_ds.patient_dice,
+    title="Dice by TKV - External dataset",
+    figureSize=figureSize,
+)
+
+
+dice_plot(
+    test_ds.TKV_GT,
+    test_ds.patient_dice,
+    title="Dice by TKV - Hold-out-test dataset",
+    figureSize=figureSize,
+)