A deep learning‐ and partial least square regression‐based model observer for a low‐contrast lesion detection task in CT

Abstract

Purpose This work aims to develop a new framework of image quality assessment using deep learning-based model observer (DL-MO) and to validate it in a low-contrast lesion detection task that involves CT images with patient anatomical background. Methods The DL-MO was developed using the transfer learning strategy to incorporate a pretrained deep convolutional neural network (CNN), a partial least square regression discriminant analysis (PLS-DA) model and an internal noise component. The CNN was previously trained to achieve the state-of-the-art classification accuracy over a natural image database. The earlier layers of the CNN were used as a deep feature extractor, with the assumption that similarity exists between the CNN and the human visual system. The PLSR model was used to further engineer the deep feature for the lesion detection task in CT images. The internal noise component was incorporated to model the inefficiency and variability of human observer (HO) performance, and to generate the ultimate DL-MO test statistics. Seven abdominal CT exams were retrospectively collected from the same type of CT scanners. To compare DL-MO with HO, 12 experimental conditions with varying lesion size, lesion contrast, radiation dose, and reconstruction types were generated, each condition with 154 trials. CT images of a real liver metastatic lesion were numerically modified to generate lesion models with four lesion sizes (5, 7, 9, and 11 mm) and three contrast levels (15, 20, and 25 HU). The lesions were inserted into patient liver images using a projection-based method. A validated noise insertion tool was used to synthesize CT exams with 50% and 25% of routine radiation dose level. CT images were reconstructed using the weighted filtered back projection algorithm and an iterative reconstruction algorithm. Four medical physicists performed a two-alternative forced choice (2AFC) detection task (with multislice scrolling viewing mode) on patient images across the 12 experimental conditions. DL-MO was operated on the same datasets. Statistical analyses were performed to evaluate the correlation and agreement between DL-MO and HO. Results A statistically significant positive correlation was observed between DL-MO and HO for the 2AFC low-contrast detection task that involves patient liver background. The corresponding Pearson product moment correlation coefficient was 0.986 [95% confidence interval (0.950, 0.996)]. Bland–Altman agreement analysis did not indicate statistically significant differences. Conclusions The proposed DL-MO is highly correlated with HO in a low-contrast detection task that involves realistic patient liver background. This study demonstrated the potential of the proposed DL-MO to assess image quality directly based on patient images in realistic, clinically relevant CT tasks.

A Convolutional Neural Network-Based Anthropomorphic Model Observer for Signal Detection in Breast CT Images Without Human-Labeled Data

“…However, since CNNs require a large amount of labeled training data [15], this approach involves extensive human observer studies, contradicting to the original purpose of anthropomorphic model observers. Another recent approach incorporates the transfer learning and an internal noise component to reduce the required amount of labeled training data and calibrate model observer performance, respectively [20]. However, the explicit use of the internal noise component might restrict the generalization performance.…”

Low‐dose CT image and projection dataset

“…As additional examples of the value of these data, within our own research program and clinical practice, we have used these and other data to. determine optimal protocol settings in our large subspecialty clinical practice, 35–38 conduct multireader, multicase (MRMC) studies to discern the impact of different reconstruction algorithms, patient dose levels and other factors on radiologist diagnostic performance and confidence, 12–16,20,35–37,39,40 develop, and evaluate using MRMC studies, nonlocal means and deep learning‐based image denoising methods, 41–43 and develop model observers and deep learning methods from phantom or patient data to predict human observer performance of radiologists when interpreting patient data to allow rapid optimization of protocols for any scanner model, exam type, or patient characteristics 17–19 …”

“…develop, and evaluate using MRMC studies, nonlocal means and deep learning-based image denoising methods, [41][42][43] and 4. develop model observers and deep learning methods from phantom or patient data to predict human observer performance of radiologists when interpreting patient data to allow rapid optimization of protocols for any scanner model, exam type, or patient characteristics. [17][18][19] This data library, however, does have several limitations. The DICOM-CT-PD format is an extended DICOM format because its header needed to contain data in private tags beyond those defined in the standard DICOM information object definition.…”

Task-based assessment of neck CT protocols using patient-mimicking phantoms—effects of protocol parameters on dose and diagnostic performance

“…Mean area under the curve values and 95% confidence intervals across all protocols and readers are shown TCM SD of 7.5 TCM SD of 14 120 kVp 0.821 (0.802 to 0.840) 0.776 (0.757 to 0.795) p = 0.003 100 kVp 0.839 (0.820 to 0.858) 0.819 (0.800 to 0.837) p = 0.354 p = 0.184 p = 0.002 performance of radiologists in the clinical setting but also subject to significant variability and time-consuming. Future work could address this limitation by using a model observer approach [33]. CT protocols vary considerably between scanners and institutions and it is likely that many patients could be examined more efficiently.…”