An interpretable deep learning approach for lesion detection and segmentation on whole-body [18F]FDG PET/CT
0
Citation
13
Reference
10
Related Paper
Abstract:
Automated lesion segmentation is essential to provide fast, reproducible tumor load estimates. Though deep learning methods have achieved unprecedented results in this field, they are often difficult to interpret, hampering their potential integration in the clinic. An interpretable deep learning approach is proposed for segmenting melanoma lesions on whole-body fluorine-18 fluorodeoxyglucose ([18F]FDG) positron emission tomography (PET) / computed tomography (CT). This consists of an automated PET thresholding step to identify FDGavid regions, followed by a three-channel nnU-Net considering the binary mask in addition to the PET and CT images. This segmentation step differentiates healthy from malignant tissue and removes the restriction on lesion boundaries imposed by the thresholding. The proposed method, trained on 267 images and evaluated on two sets acquired at the same institute, achieved mean Dice similarity coefficients (DSC) of 0.779 and 0.638 with mean absolute volume differences of 15.2mL and 22.0 mL. The DSC proved significantly higher compared to a direct, two-channel nnU-Net considering only the PET and CT. The same was observed when retraining and testing on subsets of the public data of the autoPET challenge, containing melanoma, lung cancer and lymphoma patients. In addition, overall results proved superior to a previously proposed two-step approach, where a classification network categorized each component of increased tracer uptake as healthy or malignant. The proposed lesion segmentation method for whole-body [18F]FDG PET/CT incorporates prior thresholding information while allowing more flexibility in the lesion delineation than a pure thresholding approach and increased interpretability over a direct segmentation network.In the current commercial automotive market, the need for intelligent headlight control systems has increased more and more. Camera-based night-time vehicle detection has become a crucial issue in determining the performance of such control systems. The purpose of this paper is to offer an answer to the question, 'Which thresholding method is suitable in terms of detection performance for a night-time vehicle candidate selection process?' For such purposes, two local adaptive thresholding methods are introduced and tested. One is local maximum-based thresholding, and the other is local mean-based thresholding. Efficient implementation methodologies are also introduced for real-time processing. Through the simulations tested on road image sequences with different exposure times, we prove that local adaptive thresholding methods have better performance than other well-known global thresholding methods. In particular, the simulations show that the proposed mean-based thresholding method has better performance on both long- and short-exposure sequences.
Balanced histogram thresholding
Cite
Citations (0)
Balanced histogram thresholding
Cite
Citations (0)
In this work an extension of an iterative technique for the thresholding of a given image using a single-level threshold to multi-level thresholding is presented. The technique is applied on images of machine parts and the features are extracted by border-following. The results of the multi-level thresholding and a comparison with results obtained by single-level thresholding are also presented.
Balanced histogram thresholding
Feature (linguistics)
Cite
Citations (1)
This work aimed to find a robust thresholding technique to image binarization for the gray level images. Thresholding is a simple method that plays a vital role in image segmentation. This comparative study provides to select the robust thresholding technique for general images and MRI head scans. This paper analyses the five thresholding techniques such as Sauvola thresholding, Niblack thresholding, Ridler and Calvard thresholding, Kittler and Illingworth thresholding and Otsu Thresholding for general gray images, normal and abnormal MRI head scans. The performance analysis was carried out by using the region non-uniformity parameter. Experiments were done using the mixture of gray images chosen form popularly available image databases.
Balanced histogram thresholding
Otsu's method
Gray (unit)
Cite
Citations (10)
A localized thresholding method based on boundary detection is proposed. The performance of global thresholding is likely to degrade when the gray level of the background in an image fluctuates and non-background regions with relatively large area exists. The proposed approach, localized thresholding based on boundary (LOTBOB), estimates the regions which are likely to contain the interesting objects by presuming that the appearance of boundaries indicates the existence of objects with a high possibility. Thresholding is then limited in these regions to avoid global thresholding. Experiments on real-world images show that LOTBOB is effective with relatively small and clearly-edged objects.
Balanced histogram thresholding
Cite
Citations (0)
In the current commercial automotive market, the need for intelligent headlight control systems has increased more and more. Camera-based night-time vehicle detection has become a crucial issue in determining the performance of such control systems. The purpose of this paper is to offer an answer to the question, 'Which thresholding method is suitable in terms of detection performance for a night-time vehicle candidate selection process?' For such purposes, two local adaptive thresholding methods are introduced and tested. One is local maximum-based thresholding, and the other is local mean-based thresholding. Efficient implementation methodologies are also introduced for real-time processing. Through the simulations tested on road image sequences with different exposure times, we prove that local adaptive thresholding methods have better performance than other well-known global thresholding methods. In particular, the simulations show that the proposed mean-based thresholding method has better performance on both long- and short-exposure sequences.
Balanced histogram thresholding
Cite
Citations (0)
Image thresholding is a challenging task due to its sensitivity to environmental variations and degradation in the quality of the captured image. Although many image thresholding methods have been introduced, most of them require the fine tuning of a thresholding parameter that is not suitable for single-shot structured light (SSSL) based projector-camera applications. In this paper, we introduce a locally adaptive thresholding method with automatic parameter selection based on the local statistics of the distinct image partitions. For assessing the proposed scheme, we introduce an evaluation that relies on mapping SSSL patterns between the camera and projector spaces. Experimental results demonstrate the effectiveness of the proposed technique by maintaining the thresholding accuracy of the baseline method, without the need to fine tune the obtained thresholding parameter or any noticeable change in the qualitative results.
Balanced histogram thresholding
Cite
Citations (2)
We present a variation on classic beam thresholding techniques that is up to an order of magnitude faster than the traditional method, at the same performance level. We also present a new thresholding technique, global thresholding, which, combined with the new beam thresholding, gives an additional factor of two improvement, and a novel technique, multiple pass parsing, that can be combined with the others to yield yet another 50% improvement. We use a new search algorithm to simultaneously optimize the thresholding parameters of the various algorithms.
Balanced histogram thresholding
Factor (programming language)
Cite
Citations (48)
Balanced histogram thresholding
Cite
Citations (36)