A Comparison of Transfer Learning Techniques, Deep Convolutional Neural Network and Multilayer Neural Network Methods for the Diagnosis of Glaucomatous Optic Neuropathy

Norouzifard, Mohammad; Nemati, Ali; Abdul-Rahman, Anmar; GholamHosseini, Hamid; Klette, Reinhard

doi:10.1007/978-981-13-9190-3_69

Cited by 4 publications

(7 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…30,77 In computer-aided glaucoma diagnosis methods that employ DL, the first limitation has been typically addressed in literature by using TL, which was shown in several works to outperform freshly initialized standard CNNs. 10,50,51 As for the second indicated limitation, to the best of the author's knowledge no work has previously attempted to develop a deep network specifically designed to capture diagnostically relevant information within retinal images for glaucoma diagnosis.…”

Section: Discussionmentioning

confidence: 99%

“…RIM-ONE V2 has been commonly used in literature to test the performance of several DL computer-aided glaucoma diagnosis algorithms. 10,12,50 An advantage of using independent public datasets is that different research groups can compare the…”

Section: Datasetmentioning

confidence: 99%

“…Several later works compared the performance of their designed convolutional networks to different well-known CNN architectures both with and without TL. 10,50,51 In all these works, TL was found to result in a more superior performance compared to using non-TL approaches.…”

Section: Literature Reviewmentioning

confidence: 99%

“…On the other hand, Al‐Bander et al, 12 Cerentini et al, 48 and Ovreiu et al 49 have shown that TL can be successfully used for glaucoma detection. Several later works compared the performance of their designed convolutional networks to different well‐known CNN architectures both with and without TL 10,50,51 . In all these works, TL was found to result in a more superior performance compared to using non‐TL approaches.…”

Section: Literature Reviewmentioning

confidence: 99%

See 3 more Smart Citations

TWEEC: Computer‐aided glaucoma diagnosis from retinal images using deep learning techniques

Abdel-Hamid

2021

Int J Imaging Syst Tech

View full text Add to dashboard Cite

A novel two-branched deep convolutional (TWEEC) network is developed for computer-aided glaucoma diagnosis. The TWEEC network is designed to simultaneously extract anatomical information related to the optic disc and surrounding blood vessels which are the retinal structures most affected by glaucoma progression. The spatial retinal images and wavelet approximation subbands are compared as inputs to the proposed network. TWEEC's performance is compared to three implemented convolutional networks, one of which employs transfer learning. Experiments showed that the introduced TWEEC network achieved accuracies of 98.78% and 96.34% for the spatial and wavelet inputs, respectively, by that outperforming the other three deep networks by 8-15%. This work paves the way for the development of efficient deep learning based computer-aided glaucoma diagnosis tools. Moreover, the present study illustrates that considering specific wavelet subbands for the training of convolutional networks can result in reliable performance with the advantage of reduced overall network training time.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Datasetmentioning

confidence: 99%

Section: Literature Reviewmentioning

confidence: 99%

Section: Literature Reviewmentioning

confidence: 99%

See 2 more Smart Citations

TWEEC: Computer‐aided glaucoma diagnosis from retinal images using deep learning techniques

Abdel-Hamid

2021

Int J Imaging Syst Tech

View full text Add to dashboard Cite

show abstract

“…1 . Audio extracted mel-spectrum features were fed to the input layer in a deep MNN ( multilayer neural network ) [ 9 , 12 ]. We extracted mel-spectrum features in the frequency domain using a 4096 window size (of samples) and 512 hop length (the distance between two subsequent windows).…”

Section: Methodsmentioning

confidence: 99%

A Comparison of Approaches for Synchronizing Events in Video Streams Using Audio

Norouzifard

Nemati

Mollaee

et al. 2021

Communications in Computer and Information Science

Self Cite

View full text Add to dashboard Cite

A common scenario found in experimentation is to synchronize events, such as breaks between visual stimulus, with the video record taken of an experiment made of participants as they undertake the task. In our case, we recently synchronized a protocol of stimulus presentations shown on a laptop display, with webcam video made of participants’ (who were two year old children) facial and eye movements as they were shown trials of stimulus containing moving dots (a random dot kinematogram or RDK). The purpose was to assess eye movements in response to these RDK stimulus as a part of a potential neurological assessment for children. The video contained audio signals such as “beeps” and musical interludes that indicated the start and end of trials, thereby providing a convenient opportunity to align these audio events with the timing of known events in the video record. The process of alignment can be performed manually, but this is a tedious and time consuming task when considering, for example, large databases of videos. In this paper, we tested two alternate methods for synchronizing known audio events using: 1) a deep learning based model, and a 2) standard template matching algorithm. These methods were used to synchronize the known protocol of stimulus events in videos by processing the audio contents of the recording. The deep learning approach utilized simple mel-spectrum audio signal feature extraction, whilst we adopted a cross-correlation algorithm that detected an audio template in the time domain. We found that whilst correlation was not effective as a means of beep detection; but our machine learning-based technique was robust with 90% accuracy in the testing dataset and did not the same amount of remediation required of the correlation approach.

show abstract

Attention-Guided 3D-CNN Framework for Glaucoma Detection and Structural-Functional Association Using Volumetric Images

George

Antony

Ishikawa

et al. 2020

IEEE J. Biomed. Health Inform.

View full text Add to dashboard Cite

A Comparison of Transfer Learning Techniques, Deep Convolutional Neural Network and Multilayer Neural Network Methods for the Diagnosis of Glaucomatous Optic Neuropathy

Cited by 4 publications

References 17 publications

TWEEC: Computer‐aided glaucoma diagnosis from retinal images using deep learning techniques

TWEEC: Computer‐aided glaucoma diagnosis from retinal images using deep learning techniques

A Comparison of Approaches for Synchronizing Events in Video Streams Using Audio

Attention-Guided 3D-CNN Framework for Glaucoma Detection and Structural-Functional Association Using Volumetric Images

Contact Info

Product

Resources

About