Purpose
The purpose of this study was to develop a fully automated algorithm for mammographic breast density estimation using deep learning.
Method
Our algorithm used a fully convolutional network, ...which is a deep learning framework for image segmentation, to segment both the breast and the dense fibroglandular areas on mammographic images. Using the segmented breast and dense areas, our algorithm computed the breast percent density (PD), which is the faction of dense area in a breast. Our dataset included full‐field digital screening mammograms of 604 women, which included 1208 mediolateral oblique (MLO) and 1208 craniocaudal (CC) views. We allocated 455, 58, and 91 of 604 women and their exams into training, testing, and validation datasets, respectively. We established ground truth for the breast and the dense fibroglandular areas via manual segmentation and segmentation using a simple thresholding based on BI‐RADS density assessments by radiologists, respectively. Using the mammograms and ground truth, we fine‐tuned a pretrained deep learning network to train the network to segment both the breast and the fibroglandular areas. Using the validation dataset, we evaluated the performance of the proposed algorithm against radiologists’ BI‐RADS density assessments. Specifically, we conducted a correlation analysis between a BI‐RADS density assessment of a given breast and its corresponding PD estimate by the proposed algorithm. In addition, we evaluated our algorithm in terms of its ability to classify the BI‐RADS density using PD estimates, and its ability to provide consistent PD estimates for the left and the right breast and the MLO and CC views of the same women. To show the effectiveness of our algorithm, we compared the performance of our algorithm against a state of the art algorithm, laboratory for individualized breast radiodensity assessment (LIBRA).
Result
The PD estimated by our algorithm correlated well with BI‐RADS density ratings by radiologists. Pearson's rho values of our algorithm for CC view, MLO view, and CC‐MLO‐averaged were 0.81, 0.79, and 0.85, respectively, while those of LIBRA were 0.58, 0.71, and 0.69, respectively. For CC view and CC‐MLO averaged cases, the difference in rho values between the proposed algorithm and LIBRA showed statistical significance (P < 0.006). In addition, our algorithm provided reliable PD estimates for the left and the right breast (Pearson's ρ > 0.87) and for the MLO and CC views (Pearson's ρ = 0.76). However, LIBRA showed a lower Pearson's rho value (0.66) for both the left and right breasts for the CC view. In addition, our algorithm showed an excellent ability to separate each sub BI‐RADS breast density class (statistically significant, p‐values = 0.0001 or less); only one comparison pair, density 1 and density 2 in the CC view, was not statistically significant (P = 0.54). However, LIBRA failed to separate breasts in density 1 and 2 for both the CC and MLO views (P > 0.64).
Conclusion
We have developed a new deep learning based algorithm for breast density segmentation and estimation. We showed that the proposed algorithm correlated well with BI‐RADS density assessments by radiologists and outperformed an existing state of the art algorithm.
Abstract High-risk human papillomavirus (HR-HPV) is the primary carcinogen in uterine cervical carcinoma. While genotype-specific carcinogenic risks have been extensively studied in Western ...populations, data from Korean are sparse. This study evaluates the malignant potential of the three most prevalent HR-HPVs in Korea: HPV16, HPV52, and HPV58. We analyzed 230 patients who underwent cervical conization and had been tested for HPV within a year prior to the procedure, excluding those with multiple infections. This analysis was confined to patients with single HPV infections and assessed outcomes of CIN3+, which includes carcinoma in situ (CIN3) and invasive carcinoma. The incidence of invasive cervical cancer was 6.7% for HPV16, 1.7% for HPV52, and 2.0% for HPV58; however, these differences were not statistically significant ( p = 0.187). The rate of CIN3+ for HPV16, HPV52, and HPV58 were 70.6%, 51.7%, and 58.8%, respectively. Despite the small sample size, which may limit the robustness of statistical analysis, the data suggest a higher observed risk with HPV16. These findings highlight the need for vigilant clinical management tailored to specific HPV genotypes and support the implementation of a nine-valent vaccine in Korea. Physicians should be aware of these genotype-specific risks when treating patients.
The wide heterogeneity in the appearance of breast lesions and normal breast structures can confuse computerized detection algorithms. Our purpose was therefore to develop a Lesion Highlighter (LH) ...that can improve the performance of computer-aided detection algorithms for detecting breast cancer on screening mammograms.
We hypothesized that a Cycle-GAN based Lesion Remover (LR) could act as an LH, which can improve the performance of lesion detection algorithms. We used 10,310 screening mammograms from 4,832 women that included 4,942 recalled lesions (BI-RADS 0) and 5,368 normal results (BI-RADS 1). We divided the dataset into Train:Validate:Test folds with the ratios of 0.64:0.16:0.2. We segmented image patches (400 × 400 pixels) from either lesions marked by MQSA radiologists or normal tissue in mammograms. We trained a Cycle-GAN to develop two GANs, where each GAN transferred the style of one image to another. We refer to the GAN transferring the style of a lesion to normal breast tissue as the LR. We then highlighted the lesion by color-fusing the mammogram after applying the LR to its original. Using ResNet18, DenseNet201, EfficientNetV2, and Vision Transformer as backbone architectures, we trained three deep networks for each architecture, one trained on lesion highlighted mammograms (Highlighted), another trained on the original mammograms (Baseline), and Highlighted and Baseline combined (Combined). We conducted ROC analysis for the three versions of each deep network on the test set.
The Combined version of all networks achieved AUCs ranging from 0.963 to 0.974 for identifying the image with a recalled lesion from a normal breast tissue image, which was statistically improved (p-value < 0.001) over their Baseline versions with AUCs that ranged from 0.914 to 0.967.
Our results showed that a Cycle-GAN based LR is effective for enhancing lesion conspicuity and this can improve the performance of a detection algorithm.
Abstract
Immune checkpoint inhibitors (ICIs) have substantially improved the survival of cancer patients over the past several years. However, only a minority of patients respond to ICI treatment ...(~30% in solid tumors), and current ICI-response-associated biomarkers often fail to predict the ICI treatment response. Here, we present a machine learning (ML) framework that leverages network-based analyses to identify ICI treatment biomarkers (NetBio) that can make robust predictions. We curate more than 700 ICI-treated patient samples with clinical outcomes and transcriptomic data, and observe that NetBio-based predictions accurately predict ICI treatment responses in three different cancer types—melanoma, gastric cancer, and bladder cancer. Moreover, the NetBio-based prediction is superior to predictions based on other conventional ICI treatment biomarkers, such as ICI targets or tumor microenvironment-associated markers. This work presents a network-based method to effectively select immunotherapy-response-associated biomarkers that can make robust ML-based predictions for precision oncology.
The high-risk human papillomavirus (HR-HPV) has been known as the most important carcinogen in uterine cervical carcinoma. However, there is limited evidence of the malignant potential of these ...concurrent multiple infections. This study included women who had undergone cervical conization. They underwent an HPV test by cervical swab within 12 months before the surgery. They were divided into two groups: one with a single infection with HR-HPV16 and the other with concurrent multiple infections with HR-HPVs, including genotype 16. Pathologic examination classified cases as CIS+ to assess and compare the malignant potential in both groups, including carcinoma in situ (CIS) and invasive carcinoma. Of the 220 patients infected with HR-HPV16, the single infection group consisted of 120 patients (54.5%), whereas the concurrent multiple infections consisted of 100 (45.5%) patients. The rates of HSIL were significantly higher in the concurrent multiple infection group. However, the odds ratio for CIS+ did not show a significant difference between both groups (1.417, 95% CI = 0.831–2.414, p = 0.200). The malignant potential was not significantly different between concurrent multiple infections with HR-HPVs, including 16, and a single infection with 16 in Korean women.
Plants have recently received much attention as a means of producing recombinant proteins because they are easy to grow at a low cost and at a large scale. Although many plant protein expression ...systems have been developed, there remains a need for improved systems that deliver high yields of recombinant proteins. Transcription of the recombinant gene is a key step in increasing the yield of recombinant proteins. However, revealed strong promoters, terminators, and transcription factors that have been identified do not necessarily lead to high level production of recombinant proteins. Thus, in this study, a robust expression system was designed to produce high levels of recombinant protein consisting of a novel hybrid promoter, FM'M-UD, coupled with an artificial terminator, 3PRt. FM'M-UD contained fragments from three viral promoters (the promoters of
mosaic caulimovirus (MMV) full-length transcript, the MMV subgenomic transcript, and figwort mosaic virus subgenomic transcript) and two types of
-acting elements (four GAL4 binding sites and two zinc finger binding sites). The artificial terminator, 3PRt, consisted of the PINII and 35S terminators plus RB7, a matrix attachment region. The FM'M-UD promoter increased protein levels of reporters GFP, RBD : SD1 (part of S protein from SARS-CoV-2), and human interleukin-6 (hIL6) by 4-6-fold, 2-fold, and 6-fold, respectively, relative to those of the same reporters driven by the CaMV 35S promoter. Furthermore, when the FM'M-UD/3PRt expression cassette was expressed together with GAL4/TAC3d2, an artificial transcription factor that bound the GAL4 binding sites in FM'M-UD, levels of hIL6 increased by 10.7-fold, relative to those obtained from the CaMV 35S promoter plus the RD29B terminator. Thus, this novel expression system led to the production of a large amount of recombinant protein in plants.
We conducted two analyses by comparing the transferability of a traditionally transfer-learned CNN (TL) to that of a CNN fine-tuned with an unrelated set of medical images (mammograms in this study) ...first and then fine-tuned a second time using TL, which we call the cross-organ, cross-modality transfer learned (XTL) network, on 1) multiple sclerosis (MS) segmentation of brain magnetic resonance (MR) images and 2) tumor malignancy classification of multi-parametric prostate MR images. We used 2133 screening mammograms and two public challenge datasets (longitudinal MS lesion segmentation and ProstateX) as intermediate and target datasets for XTL, respectively. We used two CNN architectures as basis networks for each analysis and fine-tuned it to match the target image types (volumetric) and tasks (segmentation and classification). We evaluated the XTL networks against the traditional TL networks using Dice coefficient and AUC as figure of merits for each analysis, respectively. For the segmentation test, XTL networks outperformed TL networks in terms of Dice coefficient (Dice coefficients of 0.72 vs 0.70 - 0.71 with p-value < 0.0001 in differences). For the classification test, XTL networks (AUCs = 0.77 - 0.80) outperformed TL networks (AUC = 0.73 - 0.75). The difference in the AUCs (AUC diff = 0.045 - 0.047) was statistically significant (p-value < 0.03). We showed XTL using mammograms improves the network performance compared to traditional TL, despite the difference in image characteristics (x-ray vs. MRI and 2D vs. 3D) and imaging tasks (classification vs. segmentation for one of the tasks).
The phytohormone abscisic acid (ABA) plays crucial roles in various physiological processes, including responses to abiotic stresses, in plants. Recently, multiple ABA transporters were identified. ...The loss-of-function and gain-of-function mutants of these transporters show altered ABA sensitivity and stomata regulation, highlighting the importance of ABA transporters in ABA-mediated processes. However, how the activity of these transporters is regulated remains elusive. Here, we show that spatial regulation of ATP BINDING CASETTE G25 (ABCG25), an ABA exporter, is an important mechanism controlling its activity. ABCG25, as a soluble green fluorescent protein (sGFP) fusion, was subject to posttranslational regulation via clathrin-dependent and adaptor protein complex-2-dependent endocytosis followed by trafficking to the vacuole. The levels of sGFP:ABCG25 at the plasma membrane (PM) were regulated by abiotic stresses and exogenously applied ABA; PM-localized sGFP:ABCG25 decreased under abiotic stress conditions via activation of endocytosis in an ABA-independent manner, but increased upon application of exogenous ABA via activation of recycling from early endosomes in an ABA-dependent manner. Based on these findings, we propose that the spatial regulation of ABCG25 is an important component of the mechanism by which plants fine-tune cellular ABA levels according to cellular and environmental conditions.
Dear Editor,
Eliminating misfolded or mistargeted proteins is crucial for cell viability because these proteins accumulate as non-specific aggregates, which can be toxic to the cell (Lee et al., ...2009; Sroka et al., 2009). Previously, we have shown that in ppi2 (plastid protein import 2) mutant plants, the transcript levels of Hsc70-4 (one isoform of the Hsc70 family) and CHIP (an E3 ligase) were highly upregulated, which ultimately plays crucial roles in proteasomal degradation of unimported plastid proteins (Lee et al., 2009). We also found that, along with those of Hsc70-4 and CHIP, the transcript level of AtBAG1 (Arabidopsis thaliana Bcl2-associated athanogene 1) in the ppi2 mutant was 2.38-fold higher than that in the wild-type (Lee et al., 2009).