Literature DB >> 32535655

Generalization of deep learning models for ultra-low-count amyloid PET/MRI using transfer learning.

Kevin T Chen1, Matti Schürer2, Jiahong Ouyang3, Mary Ellen I Koran3, Guido Davidzon3, Elizabeth Mormino4, Solveig Tiepolt2, Karl-Titus Hoffmann5, Osama Sabri2, Greg Zaharchuk3, Henryk Barthel2.   

Abstract

PURPOSE: We aimed to evaluate the performance of deep learning-based generalization of ultra-low-count amyloid PET/MRI enhancement when applied to studies acquired with different scanning hardware and protocols.
METHODS: Eighty simultaneous [18F]florbetaben PET/MRI studies were acquired, split equally between two sites (site 1: Signa PET/MRI, GE Healthcare, 39 participants, 67 ± 8 years, 23 females; site 2: mMR, Siemens Healthineers, 64 ± 11 years, 23 females) with different MRI protocols. Twenty minutes of list-mode PET data (90-110 min post-injection) were reconstructed as ground-truth. Ultra-low-count data obtained from undersampling by a factor of 100 (site 1) or the first minute of PET acquisition (site 2) were reconstructed for ultra-low-dose/ultra-short-time (1% dose and 5% time, respectively) PET images. A deep convolution neural network was pre-trained with site 1 data and either (A) directly applied or (B) trained further on site 2 data using transfer learning. Networks were also trained from scratch based on (C) site 2 data or (D) all data. Certified physicians determined amyloid uptake (+/-) status for accuracy and scored the image quality. The peak signal-to-noise ratio, structural similarity, and root-mean-squared error were calculated between images and their ground-truth counterparts. Mean regional standardized uptake value ratios (SUVR, reference region: cerebellar cortex) from 37 successful site 2 FreeSurfer segmentations were analyzed.
RESULTS: All network-synthesized images had reduced noise than their ultra-low-count reconstructions. Quantitatively, image metrics improved the most using method B, where SUVRs had the least variability from the ground-truth and the highest effect size to differentiate between positive and negative images. Method A images had lower accuracy and image quality than other methods; images synthesized from methods B-D scored similarly or better than the ground-truth images.
CONCLUSIONS: Deep learning can successfully produce diagnostic amyloid PET images from short frame reconstructions. Data bias should be considered when applying pre-trained deep ultra-low-count amyloid PET/MRI networks for generalization.

Entities:  

Keywords:  Amyloid PET; Low-count imaging; PET/MRI; Transfer learning

Mesh:

Substances:

Year:  2020        PMID: 32535655      PMCID: PMC7680289          DOI: 10.1007/s00259-020-04897-6

Source DB:  PubMed          Journal:  Eur J Nucl Med Mol Imaging        ISSN: 1619-7070            Impact factor:   9.236


  30 in total

1.  Multi-Institutional Deep Learning Modeling Without Sharing Patient Data: A Feasibility Study on Brain Tumor Segmentation.

Authors:  Micah J Sheller; G Anthony Reina; Brandon Edwards; Jason Martin; Spyridon Bakas
Journal:  Brainlesion       Date:  2019-01-26

2.  Cortical surface-based analysis. I. Segmentation and surface reconstruction.

Authors:  A M Dale; B Fischl; M I Sereno
Journal:  Neuroimage       Date:  1999-02       Impact factor: 6.556

Review 3.  PET/MR in dementia and other neurodegenerative diseases.

Authors:  Henryk Barthel; Matthias L Schroeter; Karl-Titus Hoffmann; Osama Sabri
Journal:  Semin Nucl Med       Date:  2015-05       Impact factor: 4.446

4.  Towards tracer dose reduction in PET studies: Simulation of dose reduction by retrospective randomized undersampling of list-mode data.

Authors:  Sergios Gatidis; Christian Würslin; Ferdinand Seith; Jürgen F Schäfer; Christian la Fougère; Konstantin Nikolaou; Nina F Schwenzer; Holger Schmidt
Journal:  Hell J Nucl Med       Date:  2016-03-01       Impact factor: 1.102

5.  Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach.

Authors:  E R DeLong; D M DeLong; D L Clarke-Pearson
Journal:  Biometrics       Date:  1988-09       Impact factor: 2.571

6.  Limits for Reduction of Acquisition Time and Administered Activity in 18F-FDG PET Studies of Alzheimer Dementia and Frontotemporal Dementia.

Authors:  Florian Schiller; Lars Frings; Johannes Thurow; Philipp T Meyer; Michael Mix
Journal:  J Nucl Med       Date:  2019-04-26       Impact factor: 10.057

7.  Deep Auto-context Convolutional Neural Networks for Standard-Dose PET Image Estimation from Low-Dose PET/MRI.

Authors:  Lei Xiang; Yu Qiao; Dong Nie; Le An; Qian Wang; Dinggang Shen
Journal:  Neurocomputing       Date:  2017-06-29       Impact factor: 5.719

8.  Potential Clinical Applications of PET/MR Imaging in Neurodegenerative Diseases.

Authors:  Alexander Drzezga; Henryk Barthel; Satoshi Minoshima; Osama Sabri
Journal:  J Nucl Med       Date:  2014-05-12       Impact factor: 10.057

9.  The effect of 18F-florbetapir dose reduction on region-based classification of cortical amyloid deposition.

Authors:  K Herholz; R Evans; J Anton-Rodriguez; R Hinz; J C Matthews
Journal:  Eur J Nucl Med Mol Imaging       Date:  2014-07-08       Impact factor: 9.236

Review 10.  NIA-AA Research Framework: Toward a biological definition of Alzheimer's disease.

Authors:  Clifford R Jack; David A Bennett; Kaj Blennow; Maria C Carrillo; Billy Dunn; Samantha Budd Haeberlein; David M Holtzman; William Jagust; Frank Jessen; Jason Karlawish; Enchi Liu; Jose Luis Molinuevo; Thomas Montine; Creighton Phelps; Katherine P Rankin; Christopher C Rowe; Philip Scheltens; Eric Siemers; Heather M Snyder; Reisa Sperling
Journal:  Alzheimers Dement       Date:  2018-04       Impact factor: 21.566

View more
  8 in total

Review 1.  A review on AI in PET imaging.

Authors:  Keisuke Matsubara; Masanobu Ibaraki; Mitsutaka Nemoto; Hiroshi Watabe; Yuichi Kimura
Journal:  Ann Nucl Med       Date:  2022-01-14       Impact factor: 2.668

Review 2.  Artificial Intelligence-Based Image Enhancement in PET Imaging: Noise Reduction and Resolution Enhancement.

Authors:  Juan Liu; Masoud Malekzadeh; Niloufar Mirian; Tzu-An Song; Chi Liu; Joyita Dutta
Journal:  PET Clin       Date:  2021-10

3.  A personalized deep learning denoising strategy for low-count PET images.

Authors:  Qiong Liu; Hui Liu; Niloufar Mirian; Sijin Ren; Varsha Viswanath; Joel Karp; Suleman Surti; Chi Liu
Journal:  Phys Med Biol       Date:  2022-07-13       Impact factor: 4.174

4.  Automated knee cartilage segmentation for heterogeneous clinical MRI using generative adversarial networks with transfer learning.

Authors:  Mingrui Yang; Ceylan Colak; Kishore K Chundru; Sibaji Gaj; Andreas Nanavati; Morgan H Jones; Carl S Winalski; Naveen Subhas; Xiaojuan Li
Journal:  Quant Imaging Med Surg       Date:  2022-05

Review 5.  Deep learning-based image reconstruction and post-processing methods in positron emission tomography for low-dose imaging and resolution enhancement.

Authors:  Cameron Dennis Pain; Gary F Egan; Zhaolin Chen
Journal:  Eur J Nucl Med Mol Imaging       Date:  2022-03-21       Impact factor: 10.057

Review 6.  Radiomics, machine learning, and artificial intelligence-what the neuroradiologist needs to know.

Authors:  Matthias W Wagner; Khashayar Namdar; Asthik Biswas; Suranna Monah; Farzad Khalvati; Birgit B Ertl-Wagner
Journal:  Neuroradiology       Date:  2021-09-18       Impact factor: 2.804

7.  Ultra-low-dose in brain 18F-FDG PET/MRI in clinical settings.

Authors:  Marine Soret; Jacques-Antoine Maisonobe; Serge Desarnaud; Sébastien Bergeret; Valérie Causse-Lemercier; Arnaud Berenbaum; Laura Rozenblum; Marie-Odile Habert; Aurélie Kas
Journal:  Sci Rep       Date:  2022-09-12       Impact factor: 4.996

8.  New PET technologies - embracing progress and pushing the limits.

Authors:  Nicolas Aide; Charline Lasnon; Adam Kesner; Craig S Levin; Irene Buvat; Andrei Iagaru; Ken Hermann; Ramsey D Badawi; Simon R Cherry; Kevin M Bradley; Daniel R McGowan
Journal:  Eur J Nucl Med Mol Imaging       Date:  2021-06-03       Impact factor: 9.236

  8 in total

北京卡尤迪生物科技股份有限公司 © 2022-2023.