Parallel YOLO-based Model for Real-time Mitosis Counting

Yancey, Robin Elizabeth

doi:10.24132/csrn.3201.32

Cited by 2 publications

(3 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The trained VGG16 based model, which had superior performance to the ResNet101 model, is available to anyone who wishes to access it under the name DeepSplashSpotter (DSS) [23]. Compared with existing blob detection algorithms [40][41][42][43][44][45] in medical images, the DSS can be applied to blob-like object detection problems where the blobs have low-contrast boundaries (not easily distinguished from background), are small relative to the image size, and are noisy.…”

Section: Discussionmentioning

confidence: 99%

“…Deep learning techniques are often applied to blob-like object detection in medical imaging e.g. U-Net deep learning models used for small biomarkers detection in magnetic resonance and fluorescent images [40,41], RCNNs [42] for blob detection and blob size prediction in cancer diagnosis and neuroimage processing [43][44][45]. In contrast to the blob-like objects that these algorithms have previously detected such as cell nuclei, the scintillation light splashes do not have clearly defined boundaries and instead gradually decrease in intensity at their edges before becoming indistinguishable from the background.…”

Section: Rcnn Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Automatic detection of scintillation light splashes using conventional and deep learning methods

Jiang¹,

Bugby²,

Cosma³

2022

J. Inst.

View full text Add to dashboard Cite

Six methods for the automatic detection of scintillation light splashes in a portable gamma camera are compared. Each imaging frame might contain any number of light splashes (including none), and the location and size of each light splash must be identified. For real-time imaging, splashes must be identified and characterised quickly and with minimal processing overhead. The techniques are compared on their ability to accurately determine the number, position, and size of light splashes, and to reconstruct the deposited energy within each splash for a simulated data set with known ground-truths. The speed of each technique and the ease of implementation are also discussed. For accuracy in blob (light splash) localisation, a Laplacian of Gaussian approach was found to provide the most accurate estimation. However, its performance greatly relies on the appropriate tuning of preprocessing parameters prior to image analysis and the number of blobs in each frame. Deep learning approaches (Faster-RCNNs) performed significantly better than traditional algorithms in terms of predicting the size of each light splash, did not require image preprocessing and were also more stable over a range of frame occupancies. Moreover, the paper fine-tuned a VGG16 based Faster-RCNN model with the simulated data set for the scintillation light splash detection, called DeepSplashSpotter (DSS).

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Rcnn Methodsmentioning

confidence: 99%

Automatic detection of scintillation light splashes using conventional and deep learning methods

Jiang¹,

Bugby²,

Cosma³

2022

J. Inst.

View full text Add to dashboard Cite

show abstract

“…The F1score reached 0.68 for ICPR14, making it the method that attained the highest score at this stage. R. Yancey [11] proposed the use of object detection networks such as YOLO (YOLOv3, YOLOv4scale, YOLOv5, and YOLOR) to improve the accuracy of mitosis counting. In these methods, the highest scores of 0.95 and 0.96 were achieved on the ICPR12 and ICPR14 mitosis datasets.…”

Section: Related Workmentioning

confidence: 99%

A Hybridized Deep Learning Methodology for Mitosis Detection and Classification from Histopathology Images

Nemati,

Samet,

Hancer

et al. 2023

JMIDS

View full text Add to dashboard Cite

Pathologists use histopathological images to diagnose cancer, and one key step in this process is to detect and classify mitosis. Mitosis is the process of cell division, and it is essential for normal tissue growth and repair. However, abnormal mitosis can be a sign of cancer. Therefore, the ability to accurately detect and classify mitosis is crucial for cancer diagnosis. Traditionally, pathologists rely on manual methods for this task, which are labor-intensive, time-consuming, and expensive. Computer-aided diagnosis (CAD) leverages technologies like artificial intelligence, fuzzy logic, and image processing to assist pathologists in early detection and classification. This study introduces a hybridized methodology for detecting and classifying abnormal mitosis in breast histopathological images. The proposed approach comprises two stages. In the initial stage, deep learning techniques are employed for mitosis detection. Subsequently, fuzzy-based classifiers are utilized in the second stage for mitosis classification. The methodology is applied to the ICPR12 and ICPR14 mitosis datasets. Results indicate a substantial enhancement in both accuracy and reliability of mitosis detection and classification, showcasing the effectiveness of the proposed approach.

show abstract

Parallel YOLO-based Model for Real-time Mitosis Counting

Cited by 2 publications

References 18 publications

Automatic detection of scintillation light splashes using conventional and deep learning methods

Automatic detection of scintillation light splashes using conventional and deep learning methods

A Hybridized Deep Learning Methodology for Mitosis Detection and Classification from Histopathology Images

Contact Info

Product

Resources

About