A Benchmark for automatic noise measurement in clinical computed tomography

Automatic vessel attenuation measurement for quality control of contrast-enhanced CT: Validation on the portal vein

BACKGROUND

Adequate image enhancement of organs and blood vessels of interest is an important aspect of image quality in contrast-enhanced computed tomography (CT). There is a need for an objective method for evaluation of vessel contrast that can be automatically and systematically applied to large sets of CT exams.

PURPOSE

The purpose of this work was to develop a method to automatically segment and measure attenuation Hounsfield Unit (HU) in the portal vein (PV) in contrast-enhanced abdomen CT examinations.

METHODS

Input CT images were processed by a vessel enhancing filter to determine candidate PV segmentations. Multiple machine learning (ML) classifiers were evaluated for classifying a segmentation as corresponding to the PV based on segmentation shape, location, and intensity features. A public data set of 82 contrast-enhanced abdomen CT examinations was used to train the method. An optimal ML classifier was selected by training and tuning on 66 out of the 82 exams (80% training split) in the public data set. The method was evaluated in terms of segmentation classification accuracy and PV attenuation measurement accuracy, compared to manually determined ground truth, on a test set of the remaining 16 exams (20% test split) held out from public data set. The method was further evaluated on a separate, independently collected test set of 21 examinations.

RESULTS

The best classifier was found to be a random forest, with a precision of 0.892 in the held-out test set to correctly identify the PV from among the input candidate segmentations. The mean absolute error of the measured PV attenuation relative to ground truth manual measurement was 13.4 HU. On the independent test set, the overall precision decreased to 0.684. However, the PV attenuation measurement remained relatively accurate with a mean absolute error of 15.2 HU.

CONCLUSIONS

The method was shown to accurately measure PV attenuation over a large range of attenuation values, and was validated in an independently collected dataset. The method did not require time-consuming manual contouring to supervise training. The method may be applied to systematic quality control of contrast-enhanced CT examinations.

Multiplanar reconstructions of the thoracic spine in a photon counting dual-source CT scanner: comparison to EID-CT

BACKGROUND

Photon-counting detector computed tomography (PCD-CT) is a groundbreaking technology with promising results for visualization of small bone structures.

PURPOSE

To analyze the delineation of the thoracic spine in multiplanar reconstructions (MPR) on PCD-CT compared to energy-integrating detector (EID)-CT.

MATERIAL AND METHODS

Two euthanized mice were examined using different scanners: (i) 20-slice EID-CT and (ii) dual-source PCD-CT at various CTDIVol values. Readers evaluated the thoracic spine and selected series with best visualization among signal-to-noise ratio (SNR)-matched pairs.

RESULTS

SNR was significantly higher in PCD-CT reconstructions (Br68) and lower in Hr98 reconstructions compared to EID-CT. Bone detail visualization was superior in PCD-CT (especially in Hr98 reconstructions) compared to EID-CT.

CONCLUSION

MPR on a PCD-CT had a higher SNR and better bone detail visualization even at lower radiation doses compared to EID-CT. PCD-CT with bone reconstructions showed the best delineation of small bone structures and might be considered in clinical routine.

An automated technique for global noise level measurement in CT image with a conjunction of image gradient

Automated assessment of noise level in clinical computed tomography (CT) images is a crucial technique for evaluating and ensuring the quality of these images. There are various factors that can impact CT image noise, such as statistical noise, electronic noise, structure noise, texture noise, artifact noise, etc. In this study, a method was developed to measure the global noise index (GNI) in clinical CT scans due to the fluctuation of x-ray quanta. Initially, a noise map is generated by sliding a 10 × 10 pixel for calculating Hounsfield unit (HU) standard deviation and the noise map is further combined with the gradient magnitude map. By employing Boolean operation, pixels with high gradients are excluded from the noise histogram generated with the noise map. By comparing the shape of the noise histogram from this method with Christianson's tissue-type global noise measurement algorithm, it was observed that the noise histogram computed in anthropomorphic phantoms had a similar shape with a close GNI value. In patient CT images, excluding the HU deviation due the structure change demonstrated to have consistent GNI values across the entire CT scan range with high heterogeneous tissue compared to the GNI values using Christianson's tissue-type method. The proposed GNI was evaluated in phantom scans and was found to be capable of comparing scan protocols between different scanners. The variation of GNI when using different reconstruction kernels in clinical CT images demonstrated a similar relationship between noise level and kernel sharpness as observed in uniform phantom: sharper kernel resulted in noisier images. This indicated that GNI was a suitable index for estimating the noise level in clinical CT images with either a smooth or grainy appearance. The study's results suggested that the algorithm can be effectively utilized to screen the noise level for a better CT image quality control.

Deep Learning-Based Image Noise Quantification Framework for Computed Tomography

OBJECTIVE

Noise quantification is fundamental to computed tomography (CT) image quality assessment and protocol optimization. This study proposes a deep learning-based framework, Single-scan Image Local Variance EstimatoR (SILVER), for estimating the local noise level within each region of a CT image. The local noise level will be referred to as a pixel-wise noise map.

METHODS

The SILVER architecture resembled a U-Net convolutional neural network with mean-square-error loss. To generate training data, 100 replicate scans were acquired of 3 anthropomorphic phantoms (chest, head, and pelvis) using a sequential scan mode; 120,000 phantom images were allocated into training, validation, and testing data sets. Pixel-wise noise maps were calculated for the phantom data by taking the per-pixel SD from the 100 replicate scans. For training, the convolutional neural network inputs consisted of phantom CT image patches, and the training targets consisted of the corresponding calculated pixel-wise noise maps. Following training, SILVER noise maps were evaluated using phantom and patient images. For evaluation on patient images, SILVER noise maps were compared with manual noise measurements at the heart, aorta, liver, spleen, and fat.

RESULTS

When tested on phantom images, the SILVER noise map prediction closely matched the calculated noise map target (root mean square error <8 Hounsfield units). Within 10 patient examinations, SILVER noise map had an average percent error of 5% relative to manual region-of-interest measurements.

CONCLUSION

The SILVER framework enabled accurate pixel-wise noise level estimation directly from patient images. This method is widely accessible because it operates in the image domain and requires only phantom data for training.

Fully automated image quality evaluation on patient CT: Multi-vendor and multi-reconstruction study

While the recent advancements of computed tomography (CT) technology have contributed in reducing radiation dose and image noise, an objective evaluation of image quality in patient scans has not yet been established. In this study, we present a patient-specific CT image quality evaluation method that includes fully automated measurements of noise level, structure sharpness, and alteration of structure. This study used the CT images of 120 patients from four different CT scanners reconstructed with three types of algorithm: filtered back projection (FBP), vendor-specific iterative reconstruction (IR), and a vendor-agnostic deep learning model (DLM, ClariCT.AI, ClariPi Inc.). The structure coherence feature (SCF) was used to divide an image into the homogeneous (RH) and structure edge (RS) regions, which in turn were used to localize the regions of interests (ROIs) for subsequent analysis of image quality indices. The noise level was calculated by averaging the standard deviations from five randomly selected ROIs on RH, and the mean SCFs on RS was used to estimate the structure sharpness. The structure alteration was defined by the standard deviation ratio between RS and RH on the subtraction image between FBP and IR or DLM, in which lower structure alterations indicate successful noise reduction without degradation of structure details. The estimated structure sharpness showed a high correlation of 0.793 with manually measured edge slopes. Compared to FBP, IR and DLM showed 34.38% and 51.30% noise reduction, 2.87% and 0.59% lower structure sharpness, and 2.20% and -12.03% structure alteration, respectively, on an average. DLM showed statistically superior performance to IR in all three image quality metrics. This study is expected to contribute to enhance the CT protocol optimization process by allowing a high throughput and quantitative image quality evaluation during the introduction or adjustment of lower-dose CT protocol into routine practice.

Visualization of bone details in a novel photon-counting dual-source CT scanner-comparison with energy-integrating CT

Objectives

Photon-counting detector CT (PCD-CT) promises a leap in spatial resolution due to smaller detector pixel sizes than implemented in energy-integrating detector CTs (EID-CT). Our objective was to compare the visualization of smallest bone details between PCD-CT and EID-CT using a mouse as a specimen.

Materials and methods

Two euthanized mice were scanned at a 20-slice EID-CT and a dual-source PCD-CT in single-pixel mode at various CTDI_Vol values. Image noise and signal-to-noise ratio (SNR) were evaluated using repeated ROI measurements. Edge sharpness of bones was compared by the maximal slope within CT value plots along sampling lines intersecting predefined bones of the spine. Two readers evaluated bone detail visualization at four regions of the spine on a three-point Likert scale at various CTDI_Vol’s. Two radiologists selected the series with better detail visualization among each of 20 SNR-matched pairs of EID-CT and PCD-CT series.

Results

In CTDI_Vol-matched scans, PCD-CT series showed significantly lower image noise (Noise_CTDI=5 mGy: 16.27 ± 1.39 vs. 23.46 ± 0.96 HU, p < 0.01), higher SNR (SNR_CTDI=5 mGy: 20.57 ± 1.89 vs. 14.00 ± 0.66, p < 0.01), and higher edge sharpness (Edge Slope_{lumbar spine}: 981 ± 160 vs. 608 ± 146 HU/mm, p < 0.01) than EID-CT series. Two radiologists considered the delineation of bone details as feasible at consistently lower CTDI_Vol values at PCD-CT than at EID-CT. In comparison of SNR-matched reconstructions, PCD-CT series were still considered superior in almost all cases.

Conclusions

In this head-to-head comparison, PCD-CT showed superior objective and subjective image quality characteristics over EID-CT for the delineation of tiniest bone details. Even in SNR-matched pairs (acquired at different CTDI_Vol’s), PCD-CT was strongly preferred by radiologists.

Key Points

• In dose-matched scans, photon-counting detector CT series showed significantly less image noise, higher signal-to-noise ratio, and higher edge sharpness than energy-integrating detector CT series.

• Human observers considered the delineation of tiny bone details as feasible at much lower dose levels in photon-counting detector CT than in energy-integrating detector CT.

• In direct comparison of series matched for signal-to-noise ratio, photon-counting detector CT series were considered superior in almost all cases.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00330-021-08441-4.

Assessment of the global noise algorithm for automatic noise measurement in head CT examinations

PURPOSE

The global noise (GN) algorithm has been previously introduced as a method for automatic noise measurement in clinical CT images. The accuracy of the GN algorithm has been assessed in abdomen CT examinations, but not in any other body part until now. This work assesses the GN algorithm accuracy in automatic noise measurement in head CT examinations.

METHODS

A publicly available image dataset of 99 head CT examinations was used to evaluate the accuracy of the GN algorithm in comparison to reference noise values. Reference noise values were acquired using a manual noise measurement procedure. The procedure used a consistent instruction protocol and multiple observers to mitigate the influence of intra- and interobserver variation, resulting in precise reference values. Optimal GN algorithm parameter values were determined. The GN algorithm accuracy and the corresponding statistical confidence interval were determined. The GN measurements were compared across the six different scan protocols used in this dataset. The correlation of GN to patient head size was also assessed using a linear regression model, and the CT scanner's X-ray beam quality was inferred from the model fit parameters.

RESULTS

Across all head CT examinations in the dataset, the range of reference noise was 2.9-10.2 HU. A precision of ±0.33 HU was achieved in the reference noise measurements. After optimization, the GN algorithm had a RMS error 0.34 HU corresponding to a percent RMS error of 6.6%. The GN algorithm had a bias of +3.9%. Statistically significant differences in GN were detected in 11 out of the 15 different pairs of scan protocols. The GN measurements were correlated with head size with a statistically significant regression slope parameter (p < 10-7 ). The CT scanner X-ray beam quality estimated from the slope parameter was 3.5 cm water HVL (2.8-4.8 cm 95% CI).

CONCLUSION

The GN algorithm was validated for application in head CT examinations. The GN algorithm was accurate in comparison to reference manual measurement, with errors comparable to interobserver variation in manual measurement. The GN algorithm can detect noise differences in examinations performed on different scanner models or using different scan protocols. The trend in GN across patients of different head sizes closely follows that predicted by a physical model of X-ray attenuation.