Unsupervised tumor detection in Dynamic PET/CT imaging of the prostate

Rubinstein, Eldad; Salhov, Moshe; Nidam-Leshem, Meital; White, Valerie A.; Golan, Shay; Baniel, Jack; Bernstine, Hanna; Groshar, David; Averbuch, Amir

doi:10.1016/j.media.2019.04.001

Cited by 24 publications

(15 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We initially reasoned that unsupervised analysis of MRI images followed by examination of the main image cluster may help identify many of the informative images or eliminate uninformative ones. We used a state-of-the-art deep learning algorithm, ResNet-152 27 , pretrained on ImageNet to extract features from the original images (10,904 images for 135 patients with GS = 6 and GS ≥ 8) as described elsewhere 25 . This resulted in a feature vectors containing 2,048 floating point values per image.…”

Section: Resultsmentioning

confidence: 99%

See 1 more Smart Citation

Biopsy-free prediction of prostate cancer aggressiveness using deep learning and radiology imaging

Khosravi

Lysandrou

Eljalby

et al. 2019

Preprint

View full text Add to dashboard Cite

Magnetic Resonance Imaging (MRI) is routinely used to visualize the prostate gland and manage prostate cancer. The Prostate Imaging Reporting And Data System (PI-RADS) is used to evaluate the clinical risk associated with a potential tumor. However the PI-RADS score is subjective and its assessment varies between physicians. As a result, a definite diagnosis of prostate cancer requires a biopsy to obtain tissue for pathologic analysis. A prostate biopsy is an invasive procedure and is associated with complications, including hematospermia, hematuria, and rectal bleeding. We hypothesized that an Artificial Intelligence (AI) can be trained on prostate cases where both imaging and biopsy are available to distinguish aggressive prostate cancer from non-aggressive lesions using MRI imaging only, that is, without the need for a biopsy. Our computational method, named AI-biopsy, can distinguish aggressive prostate cancer from non-aggressive disease with an AUC of 0.855 and a 79.02% accuracy. We used Class Activation Maps (CAM) to highlight which regions of MRI images are being used by our algorithm for classification, and found that AI-biopsy generally focuses on the same regions that trained uro-radiolosts focus on, with a few exceptions. In conclusion, AI-biopsy provides a data-driven and reproducible way to assess cancer aggressiveness from MRI images and a personalized strategy to reduce the number of unnecessary biopsies.

show abstract

Section: Resultsmentioning

confidence: 99%

“…Random forest achieved highest performance with an AUC of 0.82 24 . Recently Rubinstein et al used an unsupervised deep learning method to detect and localize prostate tumors in PET/CT images 25 . This study demonstrated the utility of feature selection using deep learning methods in finding tumors with AUC of 0.899 25 .…”

Section: Introductionmentioning

confidence: 99%

Biopsy-free prediction of prostate cancer aggressiveness using deep learning and radiology imaging

Khosravi

Lysandrou

Eljalby

et al. 2019

Preprint

View full text Add to dashboard Cite

show abstract

“…22 Furthermore, a tracer-specific deep denoising autoencoder (DAE)-based approach was developed by Klyuzhin et al that reduces the voxel-level noise in simulated dynamic [ C]-raclopride brain PET images 23 . Rubinstein et al developed a framework to 11 detect prostate cancer using an unsupervised learning method for early detection, and localization of malignant lesions 24 . Their recommended method is able to extract features, including statistical, kinetic biological and deep features from 4D imaging data through learning using a deep stacked convolutional auto-encoder 24 .…”

Section: Accepted Articlementioning

confidence: 99%

“…Figure 4 depicts the PSNR, SSIM, and RMSE calculated on the test dataset for each predicted frame from 14 to 26. Overall, the predicted images in the earlier frames (14-20) provide higher image quality, better noise properties, and higher quantitative accuracy than the last frames (21)(22)(23)(24)(25)(26).…”

Section: Accepted Articlementioning

confidence: 99%

Fast dynamic brain PET imaging using stochastic variational prediction for recurrent frame generation

et al. 2021

View full text Add to dashboard Cite

We assess the performance of a recurrent frame generation algorithm for prediction of late frames from initial frames in dynamic brain PET imaging.Methods: Clinical dynamic 18 F-DOPA brain PET/CT studies of 46 subjects with ten folds crossvalidation were retrospectively employed. A novel stochastic adversarial video prediction model was implemented to predict the last 13 frames (25-90 min) from the initial 13 frames (0-25 min). The quantitative analysis of the predicted dynamic PET frames was performed for the test and validation dataset using established metrics. Results:The predicted dynamic images demonstrated that the model is capable of predicting the trend of change in time-varying tracer biodistribution. The Bland-Altman plots reported the lowest tracer uptake bias (-0.04) for the putamen region and the smallest variance (95% CI: -0.38, +0.14) for the cerebellum. The region-wise Patlak graphical analysis in the caudate and putamen regions for 8 subjects from the test and validation dataset showed that the average bias for and distribution volume was 4.3%, 5.1% and 4.4%, 4.2%, (p-value < 0.05), respectively. Conclusion:We have developed a novel deep learning approach for fast dynamic brain PET imaging capable of generating the last 65 min time frames from the initial 25 min frames, thus enabling significant reduction in scanning time.

show abstract

“…healthbook TIMES Oncology Hematology healthbook Times Oncology Hematology healthbook.ch March, 2021proposed algorithm generates promising results for the detection of large cancer foci on [ 11 C]Choline PET images (AUC=0.899 vs AUC=0.812 using only the mean SUV features) 50. In the study of Mortensen et al, a convolutional neural network (CNN) was trained on [ 18 F]Choline PET/CT scans obtained before radical surgery in 45 patients to assess the feasibility of a fully automated AI-based PC quantification.…”

mentioning

confidence: 99%

Artificial Intelligence Clinical Application in Prostate Cancer Molecular Imaging

2021

hbT

View full text Add to dashboard Cite

In medical imaging, artificial intelligence (AI) can be described as the ability of a system to accurately interpret and learn from external data, acquiring knowledge to achieve specific goals and tasks through flexible adaptation. This new technology holds great promise for more personalized approaches in the future. The aim of this review is to describe the current literature on AI applied to molecular imaging for prostate cancer (PC). A comprehensive search strategy was used based on SCOPUS and PubMed databases up to October 2020. From all studies published in English, we selected the most relevant that evaluated insights of AI in the assessment of PC through positron emission tomography (PET). AI may improve PC's patient care in many different fields, from the semi-automatization of tumor segmentation, through the technical aspect of image preparation, interpretation, the calculation of additional factors based on data obtained during scanning, to prognostic prediction and risk-group selection. The implementation of AI algorithms in PC molecular imaging can improve and ease the diagnostic, predictive process, and global patient care. Construction of standardized and large databases with complex information (clinical, imaging, laboratory) will be the essential step to creating and training automated diagnostic/prognostic models that can help clinicians make unbiased and faster decisions aimed at personalized healthcare.

show abstract

Unsupervised tumor detection in Dynamic PET/CT imaging of the prostate

Cited by 24 publications

References 18 publications

Biopsy-free prediction of prostate cancer aggressiveness using deep learning and radiology imaging

Biopsy-free prediction of prostate cancer aggressiveness using deep learning and radiology imaging

Fast dynamic brain PET imaging using stochastic variational prediction for recurrent frame generation

Artificial Intelligence Clinical Application in Prostate Cancer Molecular Imaging

Contact Info

Product

Resources

About