Preclinical evaluation of a novel EGFR&c-Met bispecific near infrared probe for visualization of esophageal cancer and metastatic lymph nodes

PURPOSE

This study aimed to establish a near infrared fluorescent (NIRF) probe based on an EGFR&c-Met bispecific antibody for visualization of esophageal cancer (EC) and metastatic lymph nodes (mLNs).

METHODS

EGFR and c-Met expression were assessed by immunohistochemistry. EGFR&c-Met bispecific antibody EMB01 was labeled with IRDye800cw. The binding of EMB01-IR800 was assessed by enzyme linked immunosorbent assay, flow cytometry, and immunofluorescence. Subcutaneous tumors, orthotopic tumors, and patient-derived xenograft (PDX) were established for in vivo fluorescent imaging. PDX models using lymph nodes with or without metastasis were constructed to assess the performance of EMB01-IR800 in differential diagnosis of lymph nodes.

RESULTS

The prevalence of overexpressing EGFR or c-Met was significantly higher than single marker either in EC or corresponding mLNs. The bispecific probe EMB01-IR800 was successfully synthesized, with strong binding affinity. EMB01-IR800 showed strong cellular binding to both Kyse30 (EGFR overexpressing) and OE33 (c-Met overexpressing) cells. In vivo fluorescent imaging showed prominent EMB01-IR800 uptake in either Kyse30 or OE33 subcutaneous tumors. Likewise, EMB01-IR800 exhibited superior tumor enrichment in both thoracic orthotopic esophageal squamous cell carcinoma and abdominal orthotopic esophageal adenocarcinoma models. Moreover, EMB01-IR800 produced significantly higher fluorescence in patient-derived mLNs than in benign lymph nodes.

CONCLUSION

This study demonstrated the complementary overexpression of EGFR and c-Met in EC. Compared to single-target probes, the EGFR&c-Met bispecific NIRF probe can efficiently depict heterogeneous esophageal tumors and mLNs, which greatly increased the sensitivity of tumor and mLN identification.

CT and MRI features of hepatic epithelioid haemangioendothelioma: a multi-institutional retrospective analysis of 15 cases and a literature review

OBJECTIVE

To improve the current imaging understanding of MRI or CT for hepatic epithelioid haemangioendothelioma (HEHE) to aid in its successful preoperative diagnosis.

METHODS

The imaging features of 15 patients (median age 38.6, range 20-71; 7 M/8 F) from eight institutions with pathologically confirmed HEHE were retrospectively analysed. Additionally, the CT/MR imaging features of 180 patients in 15 literature publications were collected, analysed and compared with our case series.

RESULTS

Fifteen patients underwent CT and MRI (n = 2), CT (n = 9) or MR (n = 8) scans. A total of 92.9% (13/14) of the patients were initially diagnosed with other lesions on imaging. A total of 86.7% (13/15) were multifocal. Nodules (11/15, 73.3%) were predominantly peripheral in distribution (12/15, 80.0%). Some cases were associated with hepatic capsular retraction (13/15, 86.7%), "target signs" (8/15, 53.3%) and "lollipop signs" (5/15, 33.3%). Peripheral enhancement of various shapes in the early phase with a progressive centripetal filling was the most common pattern of enhancement (12/15, 80.0%). Abnormal vascularity was seen in 50.7% (6/15) of the patients. Suspicious tumour thromboses in the inferior vena cava were seen in 3 (20.0%) of the patients. Two of the 15 patients (13.3%) had a history of smoking.

CONCLUSIONS

HEHEs have common distinctive features, including multifocal lesions that are predominantly peripheral, "target signs", "lollipop signs", hepatic capsular retraction and peripheral enhancement of various shapes in the early phase with progressive centripetal filling. Additional aggressive imaging features that may be valuable clues to the diagnosis can be identified by CT or MRI.

Generating Full-Field Digital Mammogram From Digitized Screen-Film Mammogram for Breast Cancer Screening With High-Resolution Generative Adversarial Network

Purpose

Developing deep learning algorithms for breast cancer screening is limited due to the lack of labeled full-field digital mammograms (FFDMs). Since FFDM is a new technique that rose in recent decades and replaced digitized screen-film mammograms (DFM) as the main technique for breast cancer screening, most mammogram datasets were still stored in the form of DFM. A solution for developing deep learning algorithms based on FFDM while leveraging existing labeled DFM datasets is a generative algorithm that generates FFDM from DFM. Generating high-resolution FFDM from DFM remains a challenge due to the limitations of network capacity and lacking GPU memory.

Method

In this study, we developed a deep-learning-based generative algorithm, HRGAN, to generate synthesized FFDM (SFFDM) from DFM. More importantly, our algorithm can keep the image resolution and details while using high-resolution DFM as input. Our model used FFDM and DFM for training. First, a sliding window was used to crop DFMs and FFDMs into 256 × 256 pixels patches. Second, the patches were divided into three categories (breast, background, and boundary) by breast masks. Patches from the DFM and FFDM datasets were paired as inputs for training our model where these paired patches should be sampled from the same category of the two different image sets. U-Net liked generators and modified discriminators with two-channels output, one channel for distinguishing real and SFFDMs and the other for representing a probability map for breast mask, were used in our algorithm. Last, a study was designed to evaluate the usefulness of HRGAN. A mass segmentation task and a calcification detection task were included in the study.

Results

Two public mammography datasets, the CBIS-DDSM dataset and the INbreast dataset, were included in our experiment. The CBIS-DDSM dataset includes 753 calcification cases and 891 mass cases with verified pathology information, resulting in a total of 3568 DFMs. The INbreast dataset contains a total of 410 FFDMs with annotations of masses, calcifications, asymmetries, and distortions. There were 1784 DFMs and 205 FFDM randomly selected as Dataset A. The remaining DFMs from the CBIS-DDSM dataset were selected as Dataset B. The remaining FFDMs from the INbreast dataset were selected as Dataset C. All DFMs and FFDMs were normalized to 100μm × 100μm in our experiments. A study with a mass segmentation task and a calcification detection task was performed to evaluate the usefulness of HRGAN.

Conclusions

The proposed HRGAN can generate high-resolution SFFDMs from DFMs. Extensive experiments showed the SFFDMs were able to help improve the performance of deep-learning-based algorithms for breast cancer screening on DFM when the size of the training dataset is small.

2019 Novel Coronavirus-Infected Pneumonia on CT: A Feasibility Study of Few-Shot Learning for Computerized Diagnosis of Emergency Diseases

COVID-19 is an emerging disease with transmissibility and severity. So far, there are no effective therapeutic drugs or vaccines for COVID-19. The most serious complication of COVID-19 is a type of pneumonia called 2019 novel coronavirus-infected pneumonia (NCIP) with about 4.3% mortality rate. Comparing to chest Digital Radiography (DR), it is recently reported that chest Computed Tomography (CT) is more useful to serve as the early screening and diagnosis tool for NCIP. In this study, aimed to help physicians make the diagnostic decision, we develop a machine learning (ML) approach for automated diagnosis of NCIP on chest CT. Different from most ML approaches which often require training on thousands or millions of samples, we design a few-shot learning approach, in which we combine few-shot learning with weakly supervised model training, for computerized NCIP diagnosis. A total of 824 patients are retrospectively collected from two Hospitals with IRB approval. We first use 9 patients with clinically confirmed NCIP and 20 patients without known lung diseases for training a location detector which is a multitask deep convolutional neural network (DCNN) designed to output a probability of NCIP and the segmentation of targeted lesion area. An experienced radiologist manually localizes the potential locations of NCIPs on chest CTs of 9 COVID-19 patients and interactively segments the area of the NCIP lesions as the reference standard. Then, the multitask DCNN is furtherly fine-tuned by a weakly supervised learning scheme with 291 case-level labeled samples without lesion labels. A test set of 293 patients is independently collected for evaluation. With our NCIP-Net, the test AUC is 0.91. Our system has potential to serve as the NCIP screening and diagnosis tools for the fight of COVID-19's endemic and pandemic.

Automated fibroglandular tissue segmentation in breast MRI using generative adversarial networks

Fibroglandular tissue (FGT) segmentation is a crucial step for quantitative analysis of background parenchymal enhancement (BPE) in magnetic resonance imaging (MRI), which is useful for breast cancer risk assessment. In this study, we develop an automated deep learning method based on a generative adversarial network (GAN) to identify the FGT region in MRI volumes and evaluate its impact on a specific clinical application. The GAN consists of an improved U-Net as a generator to generate FGT candidate areas and a patch deep convolutional neural network (DCNN) as a discriminator to evaluate the authenticity of the synthetic FGT region. The proposed method has two improvements compared to the classical U-Net: (1) the improved U-Net is designed to extract more features of the FGT region for a more accurate description of the FGT region; (2) a patch DCNN is designed for discriminating the authenticity of the FGT region generated by the improved U-Net, which makes the segmentation result more stable and accurate. A dataset of 100 three-dimensional (3D) bilateral breast MRI scans from 100 patients (aged 22-78 years) was used in this study with Institutional Review Board (IRB) approval. 3D hand-segmented FGT areas for all breasts were provided as a reference standard. Five-fold cross-validation was used in training and testing of the models. The Dice similarity coefficient (DSC) and Jaccard index (JI) values were evaluated to measure the segmentation accuracy. The previous method using classical U-Net was used as a baseline in this study. In the five partitions of the cross-validation set, the GAN achieved DSC and JI values of 87.0 ± 7.0% and 77.6 ± 10.1%, respectively, while the corresponding values obtained through by the baseline method were 81.1 ± 8.7% and 69.0 ± 11.3%, respectively. The proposed method is significantly superior to the previous method using U-Net. The FGT segmentation impacted the BPE quantification application in the following manner: the correlation coefficients between the quantified BPE value and BI-RADS BPE categories provided by the radiologist were 0.46 ± 0.15 (best: 0.63) based on GAN segmented FGT areas, while the corresponding correlation coefficients were 0.41 ± 0.16 (best: 0.60) based on baseline U-Net segmented FGT areas. BPE can be quantified better using the FGT areas segmented by the proposed GAN model than using the FGT areas segmented by the baseline U-Net.

Automatic segmentation of the pectoral muscle based on boundary identification and shape prediction

The purpose of this work is to identify the pectoral muscle region in mediolateral oblique (MLO) view mammograms even when the boundary is blurred or obscured. The problem is decoupled into two subproblems in our study: identifying parts of boundaries with high confidence and predicting the overall shape of the pectoral muscle. Due to the similarity in intensity and texture between pectoral muscle and gland tissue, we trained a deep neural network to distinguish them in the first subproblem. The boundary with high confidence can be obtained according to the consistency of predictions from multiple converged models. For the shape prediction problem, a generative adversarial network (GAN) is used to learn mapping from a given identified region and the breast shape to the overall pectoral muscle shape. Our method is evaluated on a mammogram dataset including 633 MLO view mammograms collected from three different datacenters. We take U-Net as our baseline model and the dataset is divided into three groups according to the performance of U-Net for evaluation. In all three groups, U-Net achieves 80.1%, 92.9%, and 98.3% in the Dice similarity coefficient, respectively, and our method achieves 85.2%, 94.8%, and 98.1% in the Dice similarity coefficient, respectively. The experiment shows that our method effectively estimates the pectoral muscle boundary, even parts of boundaries that are difficult to detect, and greatly improves the performance of segmentation in this case.