PET Image Denoising Using a Deep Neural Network Through Fine Tuning

Gong, Kuang; Guan, Jiahui; Liu, Chih-Chieh; Qi, Jinyi

doi:10.1109/trpms.2018.2877644

Cited by 172 publications

(119 citation statements)

References 31 publications

Supporting

Mentioning

114

Contrasting

Unclassified

Order By: Relevance

“…CNNs have been used to obtain standard-dose CT and PET images from low-dose data (11,12) and to enhance images by determining scatter correction parameters (13) and CNN-augmented emission-based attenuation correction (14) in PET. Recently, Gong et al used computersimulated PET images to pretrain a denoising CNN and then fine-tuned the CNN with patient data (15). The same group has also implemented a CNN in the reconstruction process for PET data (16).…”

mentioning

confidence: 99%

Denoising of Scintillation Camera Images Using a Deep Convolutional Neural Network: A Monte Carlo Simulation Approach

2019

View full text Add to dashboard Cite

Scintillation camera images contain a large amount of Poisson noise. We have investigated whether noise can be removed in whole-body bone scans using convolutional neural networks (CNNs) trained with sets of noisy and noiseless images obtained by Monte Carlo simulation. Methods: Three CNNs were generated using 3 different sets of training images: simulated bone scan images, images of a cylindric phantom with hot and cold spots, and a mix of the first two. Each training set consisted of 40,000 noiseless and noisy image pairs. The CNNs were evaluated with simulated images of a cylindric phantom and simulated bone scan images. The mean squared error between filtered and true images was used as difference metric, and the coefficient of variation was used to estimate noise reduction. The CNNs were compared with gaussian and median filters. A clinical evaluation was performed in which the ability to detect metastases for CNNand gaussian-filtered bone scans with half the number of counts was compared with standard bone scans. Results: The best CNN reduced the coefficient of variation by, on average, 92%, and the best standard filter reduced the coefficient of variation by 88%. The best CNN gave a mean squared error that was on average 68% and 20% better than the best standard filters, for the cylindric and bone scan images, respectively. The best CNNs for the cylindric phantom and bone scans were the dedicated CNNs. No significant differences in the ability to detect metastases were found between standard, CNN-, and gaussian-filtered bone scans. Conclusion: Noise can be removed efficiently regardless of noise level with little or no resolution loss. The CNN filter enables reducing the scanning time by half and still obtaining good accuracy for bone metastasis assessment.

show abstract

mentioning

confidence: 99%

Denoising of Scintillation Camera Images Using a Deep Convolutional Neural Network: A Monte Carlo Simulation Approach

2019

View full text Add to dashboard Cite

show abstract

“…Typically, only a small number of PET clinical images with only surrogate ground-truths (manual delineations) are available. In the DL literature, the use of simulated data to train DL methods has demonstrated promise (Creswell et al 2018, Gong et al 2018 and motivated our approach to use realistic simulations that model the PET imaging physics to address training-data scarcity. However, simulated tumors may not be fully representative of tumors from the patient population and may not incorporate all tumor features.…”

Section: The Modular U-net Based DL Frameworkmentioning

confidence: 99%

A physics-guided modular deep-learning based automated framework for tumor segmentation in PET

et al. 2020

View full text Add to dashboard Cite

An important need exists for reliable PET tumor-segmentation methods for tasks such as PET-based radiation-therapy planning and reliable quantification of volumetric and radiomic features. The purpose of this study was to develop an automated physics-guided deeplearning-based PET tumor-segmentation framework that addresses the challenges of limited spatial resolution, high image noise, and lack of clinical training data with ground-truth tumor boundaries in PET imaging. We propose a three-module PET-segmentation framework in the context of segmenting primary tumors in 3D 18 F-fluorodeoxyglucose (FDG)-PET images of patients with lung cancer on a per-slice basis. The first module generates PET images containing highly realistic tumors with known ground-truth using a new stochastic and physics-based approach, addressing lack of training data. The second module trains a modified U-net using these images, helping it learn the tumor-segmentation task. The third module fine-tunes this network using a small-sized clinical dataset with radiologist-defined delineations as surrogate ground-truth, helping the framework learn features potentially missed in simulated tumors. The framework's accuracy, generalizability to different scanners, sensitivity to partial volume effects (PVEs) and efficacy in reducing the number of training images were quantitatively evaluated using Dice similarity coefficient (DSC) and several other metrics. The framework yielded reliable performance in both simulated (DSC: 0.87 (95% CI: 0.86, 0.88)) and patient images (DSC: 0.73 (95% CI: 0.71, 0.76)), outperformed several widely used semi-automated approaches, accurately segmented relatively small tumors (smallest segmented cross-section was 1.83 cm 2 ), generalized across five PET scanners (DSC: 0.74 (95% CI: 0.71, 0.76)), was relatively unaffected by PVEs, and required low training data (training with data from even 30 patients yielded DSC of 0.70 (95% CI: 0.68, 0.71)). A modular deep-learning-based framework yielded reliable automated tumor delineation in FDG-PET images of patients with lung cancer using a small-sized clinical training dataset, generalized across scanners, and demonstrated ability to segment small tumors.

show abstract

“…ML wurde in der Bildrekonstruktion [26,27], der Rauschunterdrückung [28] und der Artefaktreduktion [19] [34,35], durchgeführt werden. Eine Einteilung der Modelle in Gruppen erfolgt nach der Komplexität der Modelle: (1) die "empirischen", (2) die "analytischen", (3) die "Kompartiment"-Modelle [35] und (4) die Ganzkörper-Kompartimentmodelle, die detaillierte physiologische Strukturen und Parameter enthalten.…”

Section: Quantifizierung Der Patientenspezifischen Pharmakokinetikunclassified

Möglichkeiten zur Verbesserung der Dosimetrie und Therapieplanung in der Molekularen Radiotherapie durch maschinelles Lernen

2019

View full text Add to dashboard Cite

ZusammenfassungDie Molekulare Strahlentherapie ist eine systemische oder lokoregionäre Therapie, bei der dem Patienten ein Radionuklid oder ein radioaktiv markiertes Arzneimittel verabreicht wird. Ziel ist es, die Tumorzellen abzutöten und gleichzeitig das Normalgewebe zu schonen. Die Therapieplanung umfasst deshalb die prätherapeutische Bestimmung der zu verabreichenden Aktivitätsmenge, der Substanzmenge und der zeitlichen Abfolge möglicher die Pharmakokinetik modulierender Interventionen/Injektionen.Der Einsatz der künstlichen Intelligenz und insbesondere des maschinellen Lernens (ML) sind in letzter Zeit immer mehr in den Fokus auch der Bildgebung und Therapie mit ionisierenden Strahlen gerückt.In dieser Übersichtsarbeit wird ein Verfahren der individualisierten Therapieplanung für die Molekulare Strahlentherapie anhand eines Ablaufplanes vorgestellt. Dabei werden die Möglichkeiten diskutiert, durch den Einsatz von maschinellem Lernen die Therapieplanung zu verbessern. Der Einsatz von ML besitzt ein großes Potenzial in der Dosimetrie und Behandlungsplanung in der Molekularen Strahlentherapie. Jedoch sind noch weitere Anstrengungen erforderlich, um genügend relevante Daten zu sammeln und diese mittels ML effizient zur Verbesserung der Dosimetrie und Behandlungsplanung in der Molekularen Strahlentherapie zu nutzen.

show abstract

PET Image Denoising Using a Deep Neural Network Through Fine Tuning

Cited by 172 publications

References 31 publications

Denoising of Scintillation Camera Images Using a Deep Convolutional Neural Network: A Monte Carlo Simulation Approach

Denoising of Scintillation Camera Images Using a Deep Convolutional Neural Network: A Monte Carlo Simulation Approach

A physics-guided modular deep-learning based automated framework for tumor segmentation in PET

Möglichkeiten zur Verbesserung der Dosimetrie und Therapieplanung in der Molekularen Radiotherapie durch maschinelles Lernen

Contact Info

Product

Resources

About