Deep learning based automated left ventricle segmentation and flow quantification in 4D flow cardiac MRI.

BACKGROUND: 4D flow MRI enables assessment of cardiac function and intra-cardiac blood flow dynamics from a single acquisition. However, due to the poor contrast between the chambers and surrounding tissue, quantitative analysis relies on the segmentation derived from a registered cine MRI acquisition. This requires an additional acquisition and is prone to imperfect spatial and temporal inter-scan alignment. Therefore, in this work we developed and evaluated deep learning-based methods to segment the left ventricle (LV) from 4D flow MRI directly. METHODS: We compared five deep learning-based approaches with different network structures, data pre-processing and feature fusion methods. For the data pre-processing, the 4D flow MRI data was reformatted into a stack of short-axis view slices. Two feature fusion approaches were proposed to integrate the features from magnitude and velocity images. The networks were trained and evaluated on an in-house dataset of 101 subjects with 67,567 2D images and 3030 3D volumes. The performance was evaluated using various metrics including Dice, average surface distance (ASD), end-diastolic volume (EDV), end-systolic volume (ESV), LV ejection fraction (LVEF), LV blood flow kinetic energy (KE) and LV flow components. The Monte Carlo dropout method was used to assess the confidence and to describe the uncertainty area in the segmentation results. RESULTS: Among the five models, the model combining 2D U-Net with late fusion method operating on short-axis reformatted 4D flow volumes achieved the best results with Dice of 84.52% and ASD of 3.14 mm. The best averaged absolute and relative error between manual and automated segmentation for EDV, ESV, LVEF and KE was 19.93 ml (10.39%), 17.38 ml (22.22%), 7.37% (13.93%) and 0.07 mJ (5.61%), respectively. Flow component results derived from automated segmentation showed high correlation and small average error compared to results derived from manual segmentation. CONCLUSIONS: Deep learning-based methods can achieve accurate automated LV segmentation and subsequent quantification of volumetric and hemodynamic LV parameters from 4D flow MRI without requiring an additional cine MRI acquisition.

ASPM stabilizes the NOTCH intracellular domain 1 and promotes oncogenesis by blocking FBXW7 binding in hepatocellular carcinoma cells.

Notch signaling is aberrantly activated in approximately 30% of hepatocellular carcinoma (HCC), significantly contributing to tumorigenesis and disease progression. Expression of the major Notch receptor, NOTCH1, is upregulated in HCC cells and correlates with advanced disease stages, although the molecular mechanisms underlying its overexpression remain unclear. Here, we report that expression of the intracellular domain of NOTCH1 (NICD1) is upregulated in HCC cells due to antagonism between the E3-ubiquitin ligase F-box/WD repeat-containing protein 7 (FBXW7) and the large scaffold protein abnormal spindle-like microcephaly-associated protein (ASPM) isoform 1 (ASPM-i1). Mechanistically, FBXW7-mediated polyubiquitination and the subsequent proteasomal degradation of NICD1 are hampered by the interaction of NICD1 with ASPM-i1, thereby stabilizing NICD1 and rendering HCC cells responsive to stimulation by Notch ligands. Consistently, downregulating ASPM-i1 expression reduced the protein abundance of NICD1 but not its FBXW7-binding-deficient mutant. Reinforcing the oncogenic function of this regulatory module, the forced expression of NICD1 significantly restored the tumorigenic potential of ASPM-i1-deficient HCC cells. Echoing these findings, NICD1 was found to be strongly co-expressed with ASPM-i1 in cancer cells in human HCC tissues (P < 0.001). In conclusion, our study identifies a novel Notch signaling regulatory mechanism mediated by protein-protein interaction between NICD1, FBXW7, and ASPM-i1 in HCC cells, representing a targetable vulnerability in human HCC.

Oncogenic ASPM Is a Regulatory Hub of Developmental and Stemness Signaling in Cancers.

Despite recent advances in molecularly targeted therapies and immunotherapies, the effective treatment of advanced-stage cancers remains a largely unmet clinical need. Identifying driver mechanisms of cancer aggressiveness can lay the groundwork for the development of breakthrough therapeutic strategies. Assembly factor for spindle microtubules (ASPM) was initially identified as a centrosomal protein that regulates neurogenesis and brain size. Mounting evidence has demonstrated the pleiotropic roles of ASPM in mitosis, cell-cycle progression, and DNA double-strand breaks (DSB) repair. Recently, the exon 18-preserved isoform 1 of ASPM has emerged as a critical regulator of cancer stemness and aggressiveness in various malignant tumor types. Here, we describe the domain compositions of ASPM and its transcript variants and overview their expression patterns and prognostic significance in cancers. A summary is provided of recent progress in the molecular elucidation of ASPM as a regulatory hub of development- and stemness-associated signaling pathways, such as the Wnt, Hedgehog, and Notch pathways, and of DNA DSB repair in cancer cells. The review emphasizes the potential utility of ASPM as a cancer-agnostic and pathway-informed prognostic biomarker and therapeutic target.

Modelling chloroform in indoor swimming pool air and water: the influences of internal air circulation and occupants.

The release of chloroform from water to air in an indoor swimming pool (ISP) exhibits complex physicochemical interactions among many variables, including environmental conditions, occupant activities, and geometry of the ISP. By combining the relevant variables, a structured mathematical model, the double-layer air compartment (DLAC) model, was developed to predict the level of chloroform in ISP air. A physical parameter, the indoor airflow recycle ratio (R), was incorporated into the DLAC model due to internal airflow circulation resulting in the ISP structural configuration. The theoretical R-value for a specific indoor airflow rate (vy) can be found by fitting the predicted residence time distribution (RTD) to the simulated RTD from computational fluid dynamics (CFD), showing a positive linear relationship with vy. The mechanical energies induced by occupant activities were converted into a lumped overall mass-transfer coefficient to account for the enhanced mass transfer of chloroform from the water into the air and mixing in ISP air. The DLAC model predicted that chloroform air concentrations were statistically less accurate without considering the influence of R compared with the online open-path Fourier transform infrared measurements. A novel index, the magnitude of emission (MOE) from swimmers, was linked to the level of chloroform in ISP water. The capability of the DLAC model associated with the MOE concept may facilitate upgrading the hygiene management of ISPs, including the ability to administer necessary chlorine additives in pool water and monitor the chloroform in ISP air.

Deep learning-based prediction of intra-cardiac blood flow in long-axis cine magnetic resonance imaging.

PURPOSE: We aimed to design and evaluate a deep learning-based method to automatically predict the time-varying in-plane blood flow velocity within the cardiac cavities in long-axis cine MRI, validated against 4D flow. METHODS: A convolutional neural network (CNN) was implemented, taking cine MRI as the input and the in-plane velocity derived from the 4D flow acquisition as the ground truth. The method was evaluated using velocity vector end-point error (EPE) and angle error. Additionally, the E/A ratio and diastolic function classification derived from the predicted velocities were compared to those derived from 4D flow. RESULTS: For intra-cardiac pixels with a velocity > 5 cm/s, our method achieved an EPE of 8.65 cm/s and angle error of 41.27°. For pixels with a velocity > 25 cm/s, the angle error significantly degraded to 19.26°. Although the averaged blood flow velocity prediction was under-estimated by 26.69%, the high correlation (PCC = 0.95) of global time-varying velocity and the visual evaluation demonstrate a good agreement between our prediction and 4D flow data. The E/A ratio was derived with minimal bias, but with considerable mean absolute error of 0.39 and wide limits of agreement. The diastolic function classification showed a high accuracy of 86.9%. CONCLUSION: Using a deep learning-based algorithm, intra-cardiac blood flow velocities can be predicted from long-axis cine MRI with high correlation with 4D flow derived velocities. Visualization of the derived velocities provides adjunct functional information and may potentially be used to derive the E/A ratio from conventional CMR exams.

ASPM Activates Hedgehog and Wnt Signaling to Promote Small Cell Lung Cancer Stemness and Progression.

Small cell lung cancer (SCLC) is among the most aggressive and lethal human malignancies. Most patients with SCLC who initially respond to chemotherapy develop disease relapse. Therefore, there is a pressing need to identify novel driver mechanisms of SCLC progression to unlock treatment strategies to improve patient prognosis. SCLC cells comprise subsets of cells possessing progenitor or stem cell properties, while the underlying regulatory pathways remain elusive. Here, we identified the isoform 1 of the neurogenesis-associated protein ASPM (ASPM-I1) as a prominently upregulated stemness-associated gene during the self-renewal of SCLC cells. The expression of ASPM-I1 was found to be upregulated in SCLC cells and tissues, correlated with poor patient prognosis, and indispensable for SCLC stemness and tumorigenesis. A reporter array screening identified multiple developmental signaling pathways, including Hedgehog (Hh) and Wnt pathways, whose activity in SCLC cells depended upon ASPM-I1 expression. Mechanistically, ASPM-I1 stabilized the Hh transcriptional factor GLI1 at the protein level through a unique exon-18-encoded region by competing with the E3 ligases ß-TrCP and CUL3. In parallel, ASPM-I1 sustains the transcription of the Hh pathway transmembrane regulator SMO through the Wnt-DVL3-ß-catenin signaling axis. Functional studies verified that the ASPM-I1-regulated Hh and Wnt activities significantly contributed to SCLC aggressiveness in vivo. Consistently, the expression of ASPM-I1 positively correlated with GLI1 and stemness markers in SCLC tissues. This study illuminates an ASPM-I1-mediated regulatory module that drives tumor stemness and progression in SCLC, providing an exploitable diagnostic and therapeutic target. SIGNIFICANCE: ASPM promotes SCLC stemness and aggressiveness by stabilizing the expression of GLI1, DVL3, and SMO, representing a novel regulatory hub of Hh and Wnt signaling and targetable vulnerability.

Alpha-Mangostin Reduces Pericellular Fibronectin on Suspended Tumor Cells and Therapeutically, but Not Prophylactically, Suppresses Distant Metastasis.

Major cancer deaths can be ascribed to distant metastasis to which the assembly of pericellular fibronectin (periFN) on suspended tumor cells (STCs) in the bloodstream that facilitate endothelial attachment can lead. Even though mangosteen pericarps (MP) extracts and the major component α-mangostin (α-MG) exhibit potent cancer chemopreventive properties, whether they can prophylactically and therapeutically be used as dietary nutraceuticals to prevent distant metastasis by suppressing periFN assembly on STCs within the circulation remains obscure. Immunofluorescence staining, MTT assays, flow cytometric assays, immunoblotting, and experimental metastasis mouse models were used to detect the effects of MP extracts or α-MG on periFN on STCs, tumor cell proliferation and apoptosis, the AKT activity, and tumor lung metastasis. The periFN assembly on STCs was significantly diminished upon treatments of STCs with either α-MG or MP extracts in a dose-dependent manner without inhibiting cell proliferation and viability due to increased AKT activity. Pretreatment of STCs with α-MG appeared to suppress tumor lung metastasis and prolong mouse survival rates. Oral gavage with MP extracts could therapeutically, but not prophylactically, prevent lung metastasis of STCs. We concluded that MP extracts or the major component α-MG may therapeutically serve as a potent anti-metastatic nutraceutical.

Revealing Unforeseen Diagnostic Image Features With Deep Learning by Detecting Cardiovascular Diseases From Apical 4-Chamber Ultrasounds.

Background With the increase of highly portable, wireless, and low-cost ultrasound devices and automatic ultrasound acquisition techniques, an automated interpretation method requiring only a limited set of views as input could make preliminary cardiovascular disease diagnoses more accessible. In this study, we developed a deep learning method for automated detection of impaired left ventricular (LV) function and aortic valve (AV) regurgitation from apical 4-chamber ultrasound cineloops and investigated which anatomical structures or temporal frames provided the most relevant information for the deep learning model to enable disease classification. Methods and Results Apical 4-chamber ultrasounds were extracted from 3554 echocardiograms of patients with impaired LV function (n=928), AV regurgitation (n=738), or no significant abnormalities (n=1888). Two convolutional neural networks were trained separately to classify the respective disease cases against normal cases. The overall classification accuracy of the impaired LV function detection model was 86%, and that of the AV regurgitation detection model was 83%. Feature importance analyses demonstrated that the LV myocardium and mitral valve were important for detecting impaired LV function, whereas the tip of the mitral valve anterior leaflet, during opening, was considered important for detecting AV regurgitation. Conclusions The proposed method demonstrated the feasibility of a 3-dimensional convolutional neural network approach in detection of impaired LV function and AV regurgitation using apical 4-chamber ultrasound cineloops. The current study shows that deep learning methods can exploit large training data to detect diseases in a different way than conventionally agreed on methods, and potentially reveal unforeseen diagnostic image features.

Screening of organoids derived from patients with breast cancer implicates the repressor NCOR2 in cytotoxic stress response and antitumor immunity.

Resistance to antitumor treatment contributes to patient mortality. Functional proteomic screening of organoids derived from chemotherapy-treated patients with breast cancer identified nuclear receptor corepressor 2 (NCOR2) histone deacetylase as an inhibitor of cytotoxic stress response and antitumor immunity. High NCOR2 in the tumors of patients with breast cancer predicted chemotherapy refractoriness, tumor recurrence and poor prognosis. Molecular studies revealed that NCOR2 inhibits antitumor treatment by regulating histone deacetylase 3 (HDAC3) to repress interferon regulatory factor 1 (IRF-1)-dependent gene expression and interferon (IFN) signaling. Reducing NCOR2 or impeding its epigenetic activity by modifying its interaction with HDAC3 enhanced chemotherapy responsiveness and restored antitumor immunity. An adeno-associated viral NCOR2-HDAC3 competitor potentiated chemotherapy and immune checkpoint therapy in culture and in vivo by permitting transcription of IRF-1-regulated proapoptosis and inflammatory genes to increase IFN-γ signaling. The findings illustrate the utility of patient-derived organoids for drug discovery and suggest that targeting stress and inflammatory-repressor complexes such as NCOR2-HDAC3 could overcome treatment resistance and improve the outcome of patients with cancer.

Toward automated segmentation for acute ischemic stroke using non-contrast computed tomography.

PURPOSE: Non-contrast computed tomography (NCCT) is a first-line imaging technique for determining treatment options for acute ischemic stroke (AIS). However, its poor contrast and signal-to-noise ratio limit the diagnosis accuracy for radiologists, and automated AIS lesion segmentation using NCCT also remains a challenge. In this paper, we propose R2U-RNet, a novel model for AIS lesion segmentation using NCCT. METHODS: We used an in-house retrospective NCCT dataset with 261 AIS patients with manual lesion segmentation using follow-up diffusion-weighted images. R2U-RNet is based on an R2U-Net backbone with a novel residual refinement unit. Each input image contains two image channels from separate preprocessing procedures. The proposed model incorporates multiscale focal loss to mitigate the class imbalance problem and to leverage the importance of different levels of details. A proposed noisy-label training scheme is utilized to account for uncertainties in the manual annotations. RESULTS: The proposed model outperformed several iconic segmentation models in AIS lesion segmentation using NCCT, and our ablation study demonstrated the efficacy of the proposed model. Statistical analysis of segmentation performance revealed significant effects of regional stroke occurrence and side of the stroke, suggesting the importance of region-specific information for automated segmentation, and the potential influence of the hemispheric difference in clinical data. CONCLUSION: This study demonstrated the potentials of R2U-RNet model for automated NCCT AIS lesion segmentation. The proposed model can serve as a tool for accelerating AIS diagnoses and improving the treatment quality of AIS patients.

Integrating ensemble systems biology feature selection and bimodal deep neural network for breast cancer prognosis prediction.

Breast cancer is a heterogeneous disease. To guide proper treatment decisions for each patient, robust prognostic biomarkers, which allow reliable prognosis prediction, are necessary. Gene feature selection based on microarray data is an approach to discover potential biomarkers systematically. However, standard pure-statistical feature selection approaches often fail to incorporate prior biological knowledge and select genes that lack biological insights. Besides, due to the high dimensionality and low sample size properties of microarray data, selecting robust gene features is an intrinsically challenging problem. We hence combined systems biology feature selection with ensemble learning in this study, aiming to select genes with biological insights and robust prognostic predictive power. Moreover, to capture breast cancer's complex molecular processes, we adopted a multi-gene approach to predict the prognosis status using deep learning classifiers. We found that all ensemble approaches could improve feature selection robustness, wherein the hybrid ensemble approach led to the most robust result. Among all prognosis prediction models, the bimodal deep neural network (DNN) achieved the highest test performance, further verified by survival analysis. In summary, this study demonstrated the potential of combining ensemble learning and bimodal DNN in guiding precision medicine.

Depleting RhoA/Stress Fiber-Organized Fibronectin Matrices on Tumor Cells Non-Autonomously Aggravates Fibroblast-Driven Tumor Cell Growth.

Fibronectin (FN) expressed by tumor cells has been known to be tumor suppressive but the pericellular FN (periFN) assembled on circulating tumor cells appears to evidently promote distant metastasis. Whereas the regulation of periFN assembly in suspended cells has currently been under investigation, how it is regulated in adherent tumor cells and the role of periFN in primary tumor growth remain elusive. Techniques of RNAi, plasmid transfections, immunoblotting, fluorescence/immunohistochemistry staining, cell proliferation assays, and primary tumor growth in C57BL6 mice and Fischer 344 rats were employed in this study. We found that endogenously synthesized FN in adherent tumor cells was required for periFN assembly which was aligned by RhoA-organized actin stress fiber (SF). Depleting periFN on adherent tumor cells congruently promoted in vivo tumor growth but surprisingly did not autonomously impact on in vitro tumor cell proliferation and apoptosis, suggestive of a non-autonomous role of periFN in in vivo tumor growth. We showed that the proliferative ability of shFN-expressing tumor cells was higher than shScramble cells did in the presence of fibroblasts. Altogether, these results suggested that depriving RhoA/SF-regulated periFN matrices non-autonomously promotes fibroblast-mediated tumor cell growth.

Fibronectin in Cancer: Friend or Foe.

The role of fibronectin (FN) in tumorigenesis and malignant progression has been highly controversial. Cancerous FN plays a tumor-suppressive role, whereas it is pro-metastatic and associated with poor prognosis. Interestingly, FN matrix deposited in the tumor microenvironments (TMEs) promotes tumor progression but is paradoxically related to a better prognosis. Here, we justify how FN impacts tumor transformation and subsequently metastatic progression. Next, we try to reconcile and rationalize the seemingly conflicting roles of FN in cancer and TMEs. Finally, we propose future perspectives for potential FN-based therapeutic strategies.

The Establishment of a Lung Colonization Assay for Circulating Tumor Cell Visualization in Lung Tissues.

Metastasis is the major cause of cancer death. The role of circulating tumor cells (CTCs) in promoting cancer metastasis, in which lung colonization by CTCs critically contributes to early lung metastatic processes, has been vigorously investigated. As such, animal models are the only approach that captures the full systemic process of metastasis. Given that problems occur in previous experimental designs for examining the contributions of CTCs to blood vessel extravasation, we established an in vivo lung colonization assay in which a long-term-fluorescence cell-tracer, carboxyfluorescein succinimidyl ester (CFSE), was used to label suspended tumor cells and lung perfusion was performed to clear non-specifically trapped CTCs prior to lung removal, confocal imaging, and quantification. Polymeric fibronectin (polyFN) assembled on CTC surfaces has been found to mediate lung colonization in the final establishment of metastatic tumor tissues. Here, to specifically test the requirement of polyFN assembly on CTCs for lung colonization and extravasation, we performed short term lung colonization assays in which suspended Lewis lung carcinoma cells (LLCs) stably expressing FN-shRNA (shFN) or scramble-shRNA (shScr) and pre-labeled with 20 µM of CFSE were intravenously inoculated into C57BL/6 mice. We successfully demonstrated that the abilities of shFN LLC cells to colonize the mouse lungs were significantly diminished in comparison to shScr LLC cells. Therefore, this short-term methodology may be widely applied to specifically demonstrate the ability of CTCs within the circulation to colonize the lungs.

Apoptotic and Nonapoptotic Activities of Pterostilbene against Cancer.

Cancer is a major cause of death. The outcomes of current therapeutic strategies against cancer often ironically lead to even increased mortality due to the subsequent drug resistance and to metastatic recurrence. Alternative medicines are thus urgently needed. Cumulative evidence has pointed out that pterostilbene (trans-3,5-dimethoxy-4-hydroxystilbene, PS) has excellent pharmacological benefits for the prevention and treatment for various types of cancer in their different stages of progression by evoking apoptotic or nonapoptotic anti-cancer activities. In this review article, we first update current knowledge regarding tumor progression toward accomplishment of metastasis. Subsequently, we review current literature regarding the anti-cancer activities of PS. Finally, we provide future perspectives to clinically utilize PS as novel cancer therapeutic remedies. We, therefore, conclude and propose that PS is one ideal alternative medicine to be administered in the diet as a nutritional supplement.

Pterostilbene prevents AKT-ERK axis-mediated polymerization of surface fibronectin on suspended lung cancer cells independently of apoptosis and suppresses metastasis.

BACKGROUND: Polymeric fibronectin (polyFN) assembled on suspended breast cancer cells is required for metastasis. Conceivably, drugs that target such polyFN may fight against cancer metastasis. While stilbene analogs trigger pro-apoptotic effect on attached cancer cells, whether they prevent polyFN assembly and metastasis of suspended cancer cells via an apoptosis-independent manner remains unexplored. METHODS: We depleted suspended Lewis lung carcinoma (LLC) cells of polyFN by silencing the endogenous FN expression or pterostilbene (PS) to examine whether metastasis of lung cancer cells could thus be suppressed. We investigated whether PS regulates AKT-ERK signaling axis to suppress polyFN assembly in suspended LLC cells independently of apoptosis. We tested the therapeutic effects of orally administered PS against cancer metastasis. RESULTS: Both FN-silencing and PS among the three stilbenoids indeed significantly reduced polyFN assembly and lung metastasis of suspended LLC cells in an apoptosis-independent manner. Mechanistically, PS-induced AKT phosphorylation (pAKT) and suppressed ERK phosphorylation (pERK) in suspended LLC cells, whereas pretreatment with a PI3K inhibitor, LY294002, effectively reduced pAKT, rescued pERK, and consequently reversed the PS-suppressed polyFN assembly on LLC cells; these pretreatment effects were then overturned by the ERK inhibitor U0126. Indeed, PS-suppressed lung metastasis was counteracted by LY294002, which was further overruled with U0126. Finally, we found that PS, when orally administered in experimental metastasis assays, both significantly prevented lung colonization and metastasis of LLC cells and reduced the already established tumor growth in the mouse lungs. CONCLUSIONS: PS suppressed AKT/ERK-regulated polyFN assembly on suspended LLC cells and pulmonary metastasis. PS possesses potency in both preventing and treating lung metastasis of lung cancer cells in apoptosis-independent and apoptosis-dependent manners, respectively.

Kinetics for the distribution of acrylamide in French fries, fried oil and vapour during frying of potatoes.

Kinetic analysis for the formation of acrylamide in heated foods has been typically performed using only measured data of acrylamide in foods; however, its possible loss caused by release from heated foods into fried oil and air has seldom been considered. The results obtained from the monitoring of acrylamide by frying French fries indicated that acrylamide is distributed in three phases: French fries, frying oil, and air. From the evolved gas analysis of acrylamide and the measured concentration profile of the total acrylamide amount present in these phases, the kinetic behaviour for acrylamide formation does not obey the commonly used model of two-step consecutive reactions during frying, while a lumped kinetic model was proposed for the total acrylamide amount. Moreover, a high acrylamide level in air was observed, implying that, apart from consumers of French fries, fast-food restaurant workers are potentially subject to occupational hazards from acrylamide inhalation.