Study on analytical noise propagation in convolutional neural network methods used in computed tomography imaging

Guo, Xiaoyue; Zhang, Li; Xing, Yuxiang

doi:10.1007/s41365-022-01057-3

Cited by 13 publications

(2 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Also, the mean-square-error loss was mostly used to optimize CNN parameters in training, lacking explicit and well-balanced control of bias and uncertainty. To date, only very few works in CT field have considered CNN uncertainty and / or bias, e.g., denoising using a deterministic CNN and a loss function with weighted sampled variance and bias [13], beamhardening correction through combined CNN and FBP/iterative reconstruction using model-uncertainty-weighted image blending or prior [14], and analytic approximation of noise propagation through Unet [15]. In this work, we develop a Bayesian CNN-based MD algorithm which explicitly models, penalizes and re-balances uncertainty and bias in training, to improve reliability and trustworthiness of CNN-based MD.…”

Section: Introductionmentioning

confidence: 99%

Multi-energy CT material decomposition using Bayesian deep convolutional neural network with explicit penalty of uncertainty and bias

Gong

Leng

Baffour

et al. 2023

Medical Imaging 2023: Physics of Medical Imaging

View full text Add to dashboard Cite

Convolutional neural network (CNN)-based material decomposition has the potential to improve image quality (visual appearance) and quantitative accuracy of material maps. Most methods use deterministic CNNs with mean-square-error loss to provide point-estimates of mass densities. Point estimates can be over-confident as the reliability of CNNs is frequently compromised by bias and two major uncertainties – data and model uncertainties originating from noise in inputs and train-test data dissimilarity, respectively. Also, mean-square-error lacks explicit control of uncertainty and bias. To tackle these problems, a Bayesian dual-task CNN (BDT-CNN) with explicit penalization of uncertainty and bias was developed. It is a probabilistic CNN that concurrently conducts material classification and quantification and allows for pixel-wise modeling of bias, data uncertainty, and model uncertainty. CNN was trained with images of physical and simulated tissue-mimicking inserts at varying mass densities. Hydroxyapatite (nominal density 400mg/cc) and blood (nominal density 1095mg/cc) inserts were placed in different-sized body phantoms (30 – 45cm) and used to evaluate mean-absolute-bias (MAB) in predicted mass densities across different images at routine- and half-routine-dose. Patient CT exams were collected to assess generalizability of BDT-CNN in the presence of anatomical background. Noise insertion was used to simulate patient exams at half- and quarter-routine-dose. The deterministic dual-task CNN was used as baseline. In phantoms, BDT-CNN improved consistency of insert delineation, especially edges, and reduced overall bias (average MAB for hydroxyapatite: BDT-CNN 5.4mgHA/cc, baseline 11.0mgHA/cc and blood: BDT-CNN 8.9mgBlood/cc, baseline 14.0mgBlood/cc). In patient images, BDT-CNN improved detail preservation, lesion conspicuity, and structural consistency across different dose levels.

show abstract

Section: Introductionmentioning

confidence: 99%

Multi-energy CT material decomposition using Bayesian deep convolutional neural network with explicit penalty of uncertainty and bias

Gong

Leng

Baffour

et al. 2023

Medical Imaging 2023: Physics of Medical Imaging

View full text Add to dashboard Cite

show abstract

“…Accurate assessment of CNN uncertainty and bias is critical to establish responsible and reliable deployment of CNNbased imaging techniques. Only a few recent works have explicitly considered variance and / or bias of CNN outputs in CT denoising [2][3][4], and the corresponding methods require full access to target CNN architectures and training datasets. To date, little has been done to provide systematic assessment of uncertainty and bias of CNN models used in clinical CT tasks, especially when target CNN models and training data are non-transparent (e.g., commercial deep-learning reconstruction) to the end users.…”

Section: Introductionmentioning

confidence: 99%

Patient-specific uncertainty and bias quantification of non-transparent convolutional neural network model through knowledge distillation and Bayesian deep learning

Gong

Leng

et al. 2023

Medical Imaging 2023: Physics of Medical Imaging

View full text Add to dashboard Cite

Assessing the reliability of convolutional neural network (CNN)-based CT imaging techniques is critical for reliable deployment in practice. Some evaluation methods exist but require full access to target CNN architecture and training data, something not available for proprietary or commercial algorithms. Moreover, there is a lack of systematic evaluation methods. To address these issues, we propose a patient-specific uncertainty and bias quantification (UNIQ) method that integrates knowledge distillation and Bayesian deep learning. Knowledge distillation creates a transparent CNN (“Student CNN”) to approximate the target non-transparent CNN (“Teacher CNN”). Student CNN is built as a Bayesian-deep-learning-based probabilistic CNN that, for each input, always generates statistical distribution of the corresponding outputs, and characterizes predictive mean and two major uncertainties – data and model uncertainty. UNIQ was evaluated using a low-dose CT denoising task. Patient and phantom scans with routine-dose and synthetic quarter-dose were used to create training, validation, and testing sets. To demonstrate, Unet and Resnet were used as backbones of Teacher CNN and Student CNN respectively and were trained using independent training sets. Student Resnet was qualitatively and quantitatively evaluated. The pixel-wise predictive mean, data uncertainty, and model uncertainty from Student Resnet were very similar to the counterparts from Teacher Unet (mean-absolute-error: predictive mean 1.5HU, data uncertainty 1.8HU, model uncertainty 1.3HU; mean 2D correlation coefficient: total uncertainty 0.90, data uncertainty 0.86, model uncertainty 0.83). The proposed UNIQ can potentially systematically characterize the reliability of non-transparent CNN models used in CT.

show abstract

Gray Characteristics Analysis of Strain Field of Coal and Rock Bodies Around Boreholes During Progressive Damage Based on Digital Image

Zhang

et al. 2023

Rock Mech Rock Eng

View full text Add to dashboard Cite

Study on analytical noise propagation in convolutional neural network methods used in computed tomography imaging

Cited by 13 publications

References 32 publications

Multi-energy CT material decomposition using Bayesian deep convolutional neural network with explicit penalty of uncertainty and bias

Multi-energy CT material decomposition using Bayesian deep convolutional neural network with explicit penalty of uncertainty and bias

Patient-specific uncertainty and bias quantification of non-transparent convolutional neural network model through knowledge distillation and Bayesian deep learning

Gray Characteristics Analysis of Strain Field of Coal and Rock Bodies Around Boreholes During Progressive Damage Based on Digital Image

Contact Info

Product

Resources

About