ACQUIRING DATA
SETS
3064 slices
set
-
Publicly available kaggle data
meningioma
from 233 patients, containing
and
slices),
1426
slices), glioma (
708
(
slices)
930
pituitary tumor (
Publicly available data-set (500
slices) from 63 patients.
Renaming and merging both dataset (new
data 3564 with slice thickness is 6 mm )
Choosing the models to be
implanted on the data set
U – NET segmentation
U – NET with
classification
R2U-NET
Deep Lab V3
Pre- processing
Threshold the mask to
[0,1] values only
Normalize the images
to range [0-1]
Implement the models and add branch of
dense layers for classification to the
network
Save all models with results as zip file
Evaluate the models on the test set and
visualize the result
Compare all models, create classification reports, confusion matrix plot the loss, dice coef and accuracy
DATA SETS
MODELS
MODELS
OPERATION
COMPEAR
AMONG
MODELS

BLOCK DIGRAM 2.docx

  • 1.
    ACQUIRING DATA SETS 3064 slices set - Publiclyavailable kaggle data meningioma from 233 patients, containing and slices), 1426 slices), glioma ( 708 ( slices) 930 pituitary tumor ( Publicly available data-set (500 slices) from 63 patients. Renaming and merging both dataset (new data 3564 with slice thickness is 6 mm ) Choosing the models to be implanted on the data set U – NET segmentation U – NET with classification R2U-NET Deep Lab V3 Pre- processing Threshold the mask to [0,1] values only Normalize the images to range [0-1] Implement the models and add branch of dense layers for classification to the network Save all models with results as zip file Evaluate the models on the test set and visualize the result Compare all models, create classification reports, confusion matrix plot the loss, dice coef and accuracy DATA SETS MODELS MODELS OPERATION COMPEAR AMONG MODELS