Classification of MRI Scans

Class project made by Sagar Vakkala, RA1611003040145.

This deep learning model uses transfer learning on the VGG16 model (Which is initialised on the Imagenet weights). Two dropout layers, 1 Flatten layer and the trainable sigmoidal dense layer has been put in the end for the training.

The trace log is:

E:\Github projects\Classification-of-MRI-Scans> python .\train.py
Using TensorFlow backend.
Found 216 images belonging to 2 classes.
Found 18 images belonging to 2 classes.
2019-09-24 08:44:07.671593: I T:\src\github\tensorflow\tensorflow\core\platform\cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2
2019-09-24 08:44:07.981173: I T:\src\github\tensorflow\tensorflow\core\common_runtime\gpu\gpu_device.cc:1392] Found device 0 with properties:
name: GeForce GTX 1080 major: 6 minor: 1 memoryClockRate(GHz): 1.7715
pciBusID: 0000:26:00.0
totalMemory: 8.00GiB freeMemory: 6.60GiB
2019-09-24 08:44:07.988576: I T:\src\github\tensorflow\tensorflow\core\common_runtime\gpu\gpu_device.cc:1471] Adding visible gpu devices: 0
2019-09-24 08:44:12.364509: I T:\src\github\tensorflow\tensorflow\core\common_runtime\gpu\gpu_device.cc:952] Device interconnect StreamExecutor with strength 1 edge matrix:
2019-09-24 08:44:12.368605: I T:\src\github\tensorflow\tensorflow\core\common_runtime\gpu\gpu_device.cc:958]      0
2019-09-24 08:44:12.371299: I T:\src\github\tensorflow\tensorflow\core\common_runtime\gpu\gpu_device.cc:971] 0:   N
2019-09-24 08:44:12.374915: I T:\src\github\tensorflow\tensorflow\core\common_runtime\gpu\gpu_device.cc:1084] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 6374 MB memory) -> physical GPU (device: 0, name: GeForce GTX 1080, pci bus id: 0000:26:00.0, compute capability: 6.1)
_________________________________________________________________
Layer (type)                 Output Shape              Param #
=================================================================
vgg16 (Model)                (None, 7, 7, 512)         14714688
_________________________________________________________________
dropout_1 (Dropout)          (None, 7, 7, 512)         0
_________________________________________________________________
flatten_1 (Flatten)          (None, 25088)             0
_________________________________________________________________
dropout_2 (Dropout)          (None, 25088)             0
_________________________________________________________________
dense_1 (Dense)              (None, 1)                 25089
=================================================================
Total params: 14,739,777
Trainable params: 25,089
Non-trainable params: 14,714,688
_________________________________________________________________
Epoch 1/120
50/50 [==============================] - 26s 524ms/step - loss: 3.1889 - acc: 0.6523 - val_loss: 1.3744 - val_acc: 0.8333
Epoch 2/120
50/50 [==============================] - 20s 402ms/step - loss: 1.7422 - acc: 0.8033 - val_loss: 0.5664 - val_acc: 0.8889
Epoch 3/120
50/50 [==============================] - 20s 404ms/step - loss: 1.3608 - acc: 0.8440 - val_loss: 0.5221 - val_acc: 0.9444