I am new to AI & ML and trying to develop a model to classify a set of MRI scan images of brain tumors. It has two classes(‘Healthy’ and ‘Tumor’). My data set has 4600 images (2087 for Healthy,2513 for Tumor). I use 15 epochs, The accuracy is increasing gradually. But the validation loss and validation accuracy is going up and down. I feel it is because the model is overfitting. How can I improve my model to perform well? Thank you.

This is how I split data for training and validation.

training_data, validation_data = tf.keras.utils.image_dataset_from_directory(data_dir,
                                                     batch_size=32,
                                                     image_size=(128, 128),
                                                     validation_split=.2,
                                                     subset='both',
                                                     seed=42)

Output :
Found 4514 files belonging to 2 classes.
Using 3612 files for training.
Using 902 files for validation.

This is how images looks like

Here I rescale the data

training_data = training_data.map(lambda x,y: (x/255,y))
validation_data = validation_data.map(lambda x,y: (x/255,y))

I did Data augmentation as follows

data_augmentation = tf.keras.Sequential(
[
    tf.keras.layers.RandomFlip("horizontal"),
    tf.keras.layers.RandomRotation(0.2),
    tf.keras.layers.RandomZoom(0.2),
])

This is my model

model = tf.keras.models.Sequential()
model.add(tf.keras.layers.InputLayer(shape=(128, 128, 3)))

model.add(data_augmentation)

model.add(tf.keras.layers.Conv2D(32, kernel_size=3, activation='relu')),
model.add(tf.keras.layers.BatchNormalization()),
model.add(tf.keras.layers.MaxPooling2D(2, 2)),

model.add(tf.keras.layers.Conv2D(64, kernel_size=3, activation='relu')),
model.add(tf.keras.layers.BatchNormalization()),
model.add(tf.keras.layers.MaxPooling2D(2, 2)),

model.add(tf.keras.layers.Conv2D(128, kernel_size=3, activation='relu')),
model.add(tf.keras.layers.BatchNormalization()),
model.add(tf.keras.layers.MaxPooling2D(2, 2)),

model.add(tf.keras.layers.Flatten()),
          
model.add(tf.keras.layers.Dense(128, activation='relu')),
model.add(tf.keras.layers.BatchNormalization()),
model.add(tf.keras.layers.Dropout(0.3)),

model.add(tf.keras.layers.Dense(32, activation='relu')),
model.add(tf.keras.layers.BatchNormalization()),
model.add(tf.keras.layers.Dropout(0.3)),

model.add(tf.keras.layers.Dense(1, activation='sigmoid'))