Computer-aided detection and segmentation of malignant melanoma lesions on whole-body <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" altimg="si2.svg"><mml:msup><mml:mrow/><mml:mn>18</mml:mn></mml:msup></mml:math>F-FDG PET/CT using an interpretable deep learning approach

Dirks, Ine; Keyaerts, Marleen; Neyns, Bart; Vandemeulebroucke, Jef

doi:10.1016/j.cmpb.2022.106902

Cited by 14 publications

(18 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The dataset described in a previous publication 1 was employed with the same split into train, test and independent test sets. These data included a total of 69 patients that were treated for malignant melanoma at Universitair Ziekenhuis Brussel (UZB, Brussels, Belgium).…”

Section: Datamentioning

confidence: 99%

“…The conversion of PET images to body-weight corrected standardized uptake values (SUV bw ) and the manual lesion delineations were performed as described in the publication. 1 In brief, a spherical volume of interest (VOI) of 30 mm diameter was drawn in a homogeneous part of the liver using MIM Encore (MIM Software Inc., Cleveland, USA). From the intensities in this VOI, a threshold equaling the mean plus three times the standard deviation was derived.…”

Section: Datamentioning

confidence: 99%

“…First, all areas of increased FDG uptake are identified through automated PET thresholding as proposed in a previous work. 1 In brief, the liver is segmented using a trained V-net and within this mask, the 30 mm VOI is positioned in the region with the lowest standard deviation on both modalities that does not contain voxels with a PET intensity surpassing 5 SUV bw . For further implementation details on this first step, the reader is referred to the publication.…”

Section: Lesion Segmentationmentioning

confidence: 99%

“…For further implementation details on this first step, the reader is referred to the publication. 1 Next, this obtained mask is given as extra information together with the PET and CT image to a three-channel U-Net. 3 For this, the MONAI 4 implementation of nnU-Net 5 was used with deep supervision and a combination of dice and cross-entropy loss.…”

Section: Lesion Segmentationmentioning

confidence: 99%

See 3 more Smart Citations

An interpretable deep learning approach for lesion detection and segmentation on whole-body [18F]FDG PET/CT

Dirks,

Keyaerts,

Neyns

et al. 2024

Medical Imaging 2024: Image Processing

View full text Add to dashboard Cite

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 ([ 18 F]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 15.2 mL 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 [ 18 F]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.

show abstract

Section: Datamentioning

confidence: 99%

Section: Datamentioning

confidence: 99%

Section: Lesion Segmentationmentioning

confidence: 99%

Section: Lesion Segmentationmentioning

confidence: 99%

See 2 more Smart Citations

An interpretable deep learning approach for lesion detection and segmentation on whole-body [18F]FDG PET/CT

Dirks,

Keyaerts,

Neyns

et al. 2024

Medical Imaging 2024: Image Processing

View full text Add to dashboard Cite

show abstract

“…Accurate quantification and staging of tumors is the most important prognostic factor for predicting the survival of patients and for designing personalized patient management plans. [8,3] Analyzing PET/CT quantitatively by experienced medical imaging experts/radiologists is timeconsuming and error-prone. Automated quantitative analysis by deep learning algorithms to segment tumor lesions will enable accurate feature extraction, tumor staging, radiotherapy planning, and treatment response assessment.…”

mentioning

confidence: 99%

Automatic Whole Body FDG PET/CT Lesion Segmentation using Residual UNet and Adaptive Ensemble

Murugesan¹,

McCrumb²,

Brunner³

et al. 2023

Preprint

View full text Add to dashboard Cite

Multimodal Positron Emission Tomography/Computed Tomography (PET/CT) plays a key role in the diagnosis, staging, restaging, treatment response assessment, and radiotherapy planning of malignant tumors. The complementary nature of high-resolution anatomic CT and high sensitivity/specificity molecular PET imaging provides an accurate assessment of disease status. In oncology, 18-fluorodeoxyglucose (FDG) PET/CT is the most widely used method to identify and analyze metabolically active tumors. In particular, FDG uptake allows for more accurate detection of both nodal and distant forms of metastatic disease. Accurate quantification and staging of tumors is the most important prognostic factor for predicting the survival of patients and for designing personalized patient management plans. Analyzing PET/CT quantitatively by experienced medical imaging experts/radiologists is time-consuming and error-prone. Automated quantitative analysis by deep learning algorithms to segment tumor lesions will enable accurate feature extraction, tumor staging, radiotherapy planning, and treatment response assessment. The AutoPET Challenge 2022 provided an open-source platform to develop and benchmark deep learning models for automated PET lesion segmentation by providing large open-source whole-body FDG-PET/CT data. Using the multimodal PET/CT data from 900 subjects with 1014 studies provided by the AutoPET MICCAI 2022 Challenge, we applied fivefold cross-validation on residual UNETs to automatically segment lesions. We then utilized the output from adaptive ensemble highly contributive models as the final segmentation. Our method achieved a 10th ranking with a dice score of 0.5541 in the held-out test dataset (N=150 studies).

show abstract

An Architecture and a New Deep Learning Method for Head and Neck Cancer Prognosis by Analyzing Serial Positron Emission Tomography Images

Hurtado

Guzman

Muñoz

2023

Communications in Computer and Information Science

View full text Add to dashboard Cite

Computer-aided detection and segmentation of malignant melanoma lesions on whole-body 18F-FDG PET/CT using an interpretable deep learning approach

Cited by 14 publications

References 25 publications

An interpretable deep learning approach for lesion detection and segmentation on whole-body [18F]FDG PET/CT

An interpretable deep learning approach for lesion detection and segmentation on whole-body [18F]FDG PET/CT

Automatic Whole Body FDG PET/CT Lesion Segmentation using Residual UNet and Adaptive Ensemble

An Architecture and a New Deep Learning Method for Head and Neck Cancer Prognosis by Analyzing Serial Positron Emission Tomography Images

Contact Info

Product

Resources

About