Mitotic Indexer

Made for a school biology project

=================================================================
Total params: 129,570
Trainable params: 129,570
Non-trainable params: 0
_________________________________________________________________
188/188 [==============================] - 1s 4ms/step - loss: 0.1297 - accuracy: 0.9467
0.9466666579246521 <~ accuracy
0.12974701821804047 <~ loss

Goal

Final goal: Given an image of cells from a WSI, identify the mitotic figures and return a mitotic index.

Goal for this semester: Given an image of a cell, indetify if it is or isn't going through mitosis.

Process

All main code is in /kaggle-main/

datagen.py

• In datagen.py we use Marc Aubreville's implementation of fetching images & the annotations from the DICOM file & SQL file.
• Specify the agreedClass, slide, limit, and size, then fetch the cells that match that
• From this, we take each cell and take the needed data and push it into train_labels/train_images
• Data is shuffled and returned as (train_images, train_labels), (test_images, test_labels)

---

model.py

• datagen.py is now complete, model.py can now use the needed data

from datagen import generate_final_data

(train_images, train_labels), (test_images, test_labels) = generate_final_data()

• Add layers to model.py

model = keras.Sequential()
model.add(layers.Conv2D(32, (3, 3), activation="relu", input_shape=(40, 40, 3)))
model.add(layers.MaxPooling2D((2, 2)))
model.add(layers.Conv2D(64, (3, 3), activation="relu"))
model.add(layers.MaxPooling2D((2, 2)))
model.add(layers.Conv2D(64, (3, 3), activation="relu"))
model.add(layers.MaxPooling2D((2, 2)))
model.add(layers.Conv2D(64, (3, 3), activation="relu"))
model.add(layers.Flatten())
model.add(layers.Dense(64, activation="relu"))
model.add(layers.Dense(2, activation="softmax"))

• Setup model.compile()

optimizer="adam",
loss="categorical_crossentropy",
metrics=["accuracy"])

• Evaluating model

test_loss, test_acc = model.evaluate(test_images, test_labels)

print(test_acc)
print(test_loss)

• Results!

=================================================================
Total params: 129,570
Trainable params: 129,570
Non-trainable params: 0
_________________________________________________________________
188/188 [==============================] - 1s 4ms/step - loss: 0.1297 - accuracy: 0.9467
0.9466666579246521 <~ accuracy
0.12974701821804047 <~ loss