Skripsi
SEGMENTASI CITRA RETINA MENGGUNAKAN ARSITEKTUR RESIDUAL NETWORK DAN CONVOLUTIONAL LONG SHORT-TERM MEMORY
Diabetic Retinopathy is a condition that often causes blindness in the retina. One characteristic of diabetic retinopathy in the retina is the presence of exudates. Exudates segmentation is one way to detect diabetic retinopathy. A commonly used method for exudates segmentation is Convolutional Neural Network (CNN). One architecture found in CNN methods is U-Net.. Even though it is successful in some cases, U-Net sometimes still faces problems such as experiencing overfitting caused by the large number of parameters. To overcome this obstacle, it is necessary to modify the U-Net architecture. This study proposes combining the Residual Network (ResNet) architecture with Convolutional Long Short-Term Memory (ConvLSTM). The encoder part of U-Net is replaced with ResNet to reduce parameters and prevent overfitting. The use of ResNet may result in some features being skipped and not learned. To address this, the decoder part of U-Net is replaced with ConvLSTM so that it can recall the skipped features in the ResNet section. This architecture is capable of producing exudates segmentation with high accuracy. This method has the potential to improve the diagnosis of Diabetic Retinopathy with good segmentation of exudates.
Inventory Code | Barcode | Call Number | Location | Status |
---|---|---|---|---|
2407003754 | 2407003754 | 24070037542024 | Central Library (references) | Available but not for loan - Not for Loan |
No other version available