Search

We consider coupled models of flow and energy with phase change in thawing soils in permafrost at the pore- and Darcy scales. We develop the underlying models from first principles and define robust and conservative discretization algorithms based on cell-centered finite volume discretization on Cartesian grids and implicit time stepping for individual models. To couple the models we consider sequential approaches as well as time-lagging of some of the nonlinear properties. We also derive practical new surrogate models to obtain Darcy scale data from the pore-scale simulations. Finally, we study the sensitivity of models to the data in realistic scenarios. In particular we find that for the dynamics at the time scale of days, the nonlinear flow parameters have most significant impact on pressure, while temperature is less sensitive to data. [ABSTRACT FROM AUTHOR]

Background: Accurate individualized assessment of preeclampsia risk enables the identification of patients most likely to benefit from initiation of low-dose aspirin at 12 to 16 weeks of gestation when there is evidence for its effectiveness, and enables the guidance of appropriate pregnancy care pathways and surveillance., Objective: The primary objective of this study was to evaluate the performance of artificial neural network models for the prediction of preterm preeclampsia (<37 weeks' gestation) using patient characteristics available at the first antenatal visit and data from prenatal cell-free DNA screening. Secondary outcomes were prediction of early-onset preeclampsia (<34 weeks' gestation) and term preeclampsia (≥37 weeks' gestation)., Methods: This secondary analysis of a prospective, multicenter, observational prenatal cell-free DNA screening study (SMART) included singleton pregnancies with known pregnancy outcomes. Thirteen patient characteristics that are routinely collected at the first prenatal visit and 2 characteristics of cell-free DNA (total cell-free DNA and fetal fraction) were used to develop predictive models for early-onset (<34 weeks), preterm (<37 weeks), and term (≥37 weeks) preeclampsia. For the models, the "reference" classifier was a shallow logistic regression model. We also explored several feedforward (nonlinear) neural network architectures with ≥1 hidden layers, and compared their performance with the logistic regression model. We selected a simple neural network model built with 1 hidden layer and made up of 15 units., Results: Of the 17,520 participants included in the final analysis, 72 (0.4%) developed early-onset, 251 (1.4%) preterm, and 420 (2.4%) term preeclampsia. Median gestational age at cell-free DNA measurement was 12.6 weeks, and 2155 (12.3%) had their cell-free DNA measurement at ≥16 weeks' gestation. Preeclampsia was associated with higher total cell-free DNA (median, 362.3 vs 339.0 copies/mL cell-free DNA; P<.001) and lower fetal fraction (median, 7.5% vs 9.4%; P<.001). The expected, cross-validated area under the curve scores for early-onset, preterm, and term preeclampsia were 0.782, 0.801, and 0.712, respectively, for the logistic regression model, and 0.797, 0.800, and 0.713, respectively, for the neural network model. At a screen-positive rate of 15%, sensitivity for preterm preeclampsia was 58.4% (95% confidence interval, 0.569-0.599) for the logistic regression model and 59.3% (95% confidence interval, 0.578-0.608) for the neural network model. The contribution of both total cell-free DNA and fetal fraction to the prediction of term and preterm preeclampsia was negligible. For early-onset preeclampsia, removal of the total cell-free DNA and fetal fraction features from the neural network model was associated with a 6.9% decrease in sensitivity at a 15% screen-positive rate, from 54.9% (95% confidence interval, 52.9-56.9) to 48.0% (95% confidence interval, 45.0-51.0)., Conclusion: Routinely available patient characteristics and cell-free DNA markers can be used to predict preeclampsia with performance comparable to that of other patient characteristic models for the prediction of preterm preeclampsia. Logistic regression and neural network models showed similar performance., (Copyright © 2024. Published by Elsevier Inc.)

Objective: One goal of prenatal genetic screening is to optimize perinatal care and improve infant outcomes. We sought to determine whether high-risk cfDNA screening for 22q11.2 deletion syndrome (22q11.2DS) affected prenatal or neonatal management., Methods: This was a secondary analysis from the SMART study. Patients with high-risk cfDNA results for 22q11.2DS were compared with the low-risk cohort for pregnancy characteristics and obstetrical management. To assess differences in neonatal care, we compared high-risk neonates without prenatal genetic confirmation with a 1:1 matched low-risk cohort., Results: Of 18,020 eligible participants enrolled between 2015 and 2019, 38 (0.21%) were high-risk and 17,982 (99.79%) were low-risk for 22q11.2DS by cfDNA screening. High-risk participants had more prenatal diagnostic testing (55.3%; 21/38 vs. 2.0%; 352/17,982, p < 0.001) and fetal echocardiography (76.9%; 10/13 vs. 19.6%; 10/51, p < 0.001). High-risk newborns without prenatal diagnostic testing had higher rates of neonatal genetic testing (46.2%; 6/13 vs. 0%; 0/51, P < 0.001), echocardiography (30.8%; 4/13 vs. 4.0%; 2/50, p = 0.013), evaluation of calcium levels (46.2%; 6/13 vs. 4.1%; 2/49, P < 0.001) and lymphocyte count (53.8%; 7/13 vs. 15.7%; 8/51, p = 0.008)., Conclusions: High-risk screening results for 22q11.2DS were associated with higher rates of prenatal and neonatal diagnostic genetic testing and other 22q11.2DS-specific evaluations. However, these interventions were not universally performed, and >50% of high-risk infants were discharged without genetic testing, representing possible missed opportunities to improve outcomes for affected individuals., (© 2023 The Authors. Prenatal Diagnosis published by John Wiley & Sons Ltd.)

Background: The clinical implications of nonreportable cell-free DNA screening results are uncertain, but such results may indicate poor placental implantation in some cases and be associated with adverse obstetrical and perinatal outcomes., Objective: This study aimed to assess the outcomes of pregnancies with nonreportable cell-free DNA screening in a cohort of patients with complete genetic and obstetrical outcomes., Study Design: This was a prespecified secondary analysis of a multicenter prospective observational study of prenatal cell-free DNA screening for fetal aneuploidy and 22q11.2 deletion syndrome. Participants who underwent cell-free DNA screening from April 2015 through January 2019 were offered participation. Obstetrical outcomes and neonatal genetic testing results were collected from 21 primary-care and referral centers in the United States, Europe, and Australia. The primary outcome was risk for adverse obstetrical and perinatal outcomes (aneuploidy, preterm birth at <28, <34, and <37 weeks' gestation, preeclampsia, small for gestational age or birthweight <10th percentile for gestational week, and a composite outcome that included preterm birth at <37 weeks, preeclampsia, small for gestational age, and stillbirth at >20 weeks) after nonreportable cell-free DNA screening because of low fetal fraction or other causes. Multivariable analyses were performed, adjusting for variables known to be associated with obstetrical and perinatal outcomes, nonreportable results, or fetal fraction., Results: In total, 25,199 pregnant individuals were screened, and 20,194 were enrolled. Genetic confirmation was missing in 1165 (5.8%), 1085 (5.4%) were lost to follow-up, and 93 (0.5%) withdrew; the final study cohort included 17,851 (88.4%) participants who had cell-free DNA, fetal or newborn genetic confirmatory testing, and obstetrical and perinatal outcomes collected. Results were nonreportable in 602 (3.4%) participants. A sample was redrawn and testing attempted again in 427; in 112 (26.2%) participants, results were again nonreportable. Nonreportable results were associated with higher body mass index, chronic hypertension, later gestational age, lower fetal fraction, and Black race. Trisomy 13, 18, or 21 was confirmed in 1.6% with nonreportable tests vs 0.7% with reported results (P=.013). Rates of preterm birth at <28, 34, and 37 weeks, preeclampsia, and the composite outcome were higher among participants with nonreportable results, and further increased among those with a second nonreportable test, whereas the rate of small for gestational age infants was not increased. After adjustment for confounders, the adjusted odds ratios were 2.2 (95% confidence interval, 1.1-4.4) and 2.6 (95% confidence interval, 0.6-10.8) for aneuploidy, and 1.5 (95% confidence interval, 1.2-1.8) and 2.1 (95% confidence interval, 1.4-3.2) for the composite outcome after a first and second nonreportable test, respectively. Of the patients with nonreportable tests, 94.9% had a live birth, as opposed to 98.8% of those with reported test results (adjusted odds ratio for livebirth, 0.20 [95% confidence interval, 0.13-0.30])., Conclusion: Patients with nonreportable cell-free DNA results are at increased risk for a number of adverse outcomes, including aneuploidy, preeclampsia, and preterm birth. They should be offered diagnostic genetic testing, and clinicians should be aware of the increased risk of pregnancy complications., (Copyright © 2023 The Authors. Published by Elsevier Inc. All rights reserved.)

Background: It is estimated that approximately 300,000 people are experiencing homelessness in England. The aim of this study was to evaluate key causes and long-term trends of emergency departments (EDs) and in hospital inpatient admissions of persons experiencing homelessness in England., Methods: ED and hospital inpatient admissions data were obtained from Hospital Episode Statistics (HES) covering all National Health Service (NHS) England hospitals. Anyone identified or declared to be experiencing homelessness during the service usage are recorded in HES datasets. Data were extracted for the 10-year study period and compared to the general population, which includes all patients attending the ED or admitted to inpatient care in England., Results: Drug- and alcohol-related causes contribute to the most frequent reasons for attendance and admissions of persons experiencing homelessness in the ED and inpatient respectively. A total of 30,406 ED attendances were recorded for persons experiencing homelessness in the year 2018/2019 (+ 44.9% rise vs 2009/10) of which injuries and poisoning respectively represented 21.8% and 17.9% of all persons experiencing homelessness presentations to the ED. Poisoning (including drug overdose) represented only 1.9% of all attendances by the general population during the same study year (rate ratio vs general populations 9.2 95% CI 9.0-9.4). High mortality rates were observed in relation to presentations attributed to drug- and alcohol-related causes. A total of 14,858 persons experiencing homelessness inpatient admissions were recorded in 2018/2019 (+ 68.6% vs 2009/2010). Psychoactive substance use constituted 12.7% of all admissions in 2018/2019 compared to 0.4% of in the general populations (rate ratio: 33.3, 95% CI: 31.9-34.7). There was a 44.3% rise in the number of admissions related to poisoning in the study period amongst persons experiencing homelessness in England (vs 14.2% in general population)., Conclusion: Marked disparities around primary causes of ED and inpatient admissions were identified between persons experiencing homelessness and the general population. There is a continued need for prevention measures to reduce the prevalence of drug and alcohol, injury and poisoning-related admissions to the ED, enhanced service provision at the community level, and multisector collaborations. These initiatives should maximise opportunities for early interventions and improve outcomes for persons experiencing homelessness, including increased accessibility of healthcare and mental health services, particularly in areas that demonstrate increasing ED and inpatient attendance rates over time., (© 2023. Springer-Verlag GmbH Germany, part of Springer Nature.)

Purpose: The aim of this study was to assess the performance of cell-free DNA (cfDNA) screening to detect sex chromosome aneuploidies (SCAs) in an unselected obstetrical population with genetic confirmation., Methods: This was a planned secondary analysis of the multicenter, prospective SNP-based Microdeletion and Aneuploidy RegisTry (SMART) study. Patients receiving cfDNA results for autosomal aneuploidies and who had confirmatory genetic results for the relevant sex chromosomal aneuploidies were included. Screening performance for SCAs, including monosomy X (MX) and the sex chromosome trisomies (SCT: 47,XXX; 47,XXY; 47,XYY) was determined. Fetal sex concordance between cfDNA and genetic screening was also evaluated in euploid pregnancies., Results: A total of 17,538 cases met inclusion criteria. Performance of cfDNA for MX, SCTs, and fetal sex was determined in 17,297, 10,333, and 14,486 pregnancies, respectively. Sensitivity, specificity, and positive predictive value (PPV) of cfDNA were 83.3%, 99.9%, and 22.7% for MX and 70.4%, 99.9%, and 82.6%, respectively, for the combined SCTs. The accuracy of fetal sex prediction by cfDNA was 100%., Conclusion: Screening performance of cfDNA for SCAs is comparable to that reported in other studies. The PPV for the SCTs was similar to the autosomal trisomies, whereas the PPV for MX was substantially lower. No discordance in fetal sex was observed between cfDNA and postnatal genetic screening in euploid pregnancies. These data will assist interpretation and counseling for cfDNA results for sex chromosomes., Competing Interests: Conflict of Interest All site principal investigators (Pe’er Dar, Bo Jacobsson, Fergal Malone, Ronald J. Wapner, Ashley S. Roman, Asma Khalil, Revital Faro, Rajeevi Madankumar, Sina Haeri, Robert M. Silver, Nidhi Vohra, Jon Hyett, Cora MacPherson, and Mary E. Norton) received institutional research support from the funding sponsor (Natera). Melissa Egbert, Zachary Demko, and Matt Rabinowitz are employed by the study’s funding sponsor (Natera) and hold stock or options to hold stock. Kimberly Martin is a consultant to the funding sponsor (Natera) and holds stock and options to hold stock. Jon Hyett has an ongoing research collaboration that includes financial support for biochemical analytes from Perkin Elmer; has earned honoraria and/or given talks that were not compensated from Natera, Roche, and Canon; and has participated in Asian/Australasian expert consultancies for Natera and Roche. Bo Jacobsson collaborated in the Improving Maternal Pregnancy And Child ouTcomes (IMPACT) study where Roche, Perkin Elmer, and Thermo Fisher provide reagents to Placental Growth Factor (PLGF) analyses. Ronald J. Wapner receives research funding from National Institute of Child Health and Human Development and support from Illumina for research reagents. Mary E. Norton is a consultant to Luna Genetics. All other authors declare no conflicts of interest., (Copyright © 2023 The Authors. Published by Elsevier Inc. All rights reserved.)

Semantic Artificial Intelligence has certain qualities that are advantageous for deep learning-based medical imaging tasks. Medical images can be augmented by injecting semantic context into the underlying classification mechanism, increasing the information density of the scan and ultimately can provide more trust in the result. This work considers an application of semantic AI to segment tissue types from excised breast tumors imaged with pulsed terahertz (THz)-an emerging imaging technology. Prior work has demonstrated traditional data driven methodology for deep learning on THz has two key weaknesses: namely 1) low image resolution compared to other state-of-the-art imaging techniques and 2) a lack of expertly-labeled images due to domain transformation and tissue changes during histopathology. This work seeks to address these limitations through two semantic AI mechanisms. Specifically, we introduce a two stage pipeline using an unsupervised image-to-image translation network and a supervised segmentation network. The combination of these contributions enables enhanced near-real-time visualization of excised tissue through THz scans and a supervised segmentation and classification training strategy using only synthetic THz scans generated by our bi-directional image-to-image translation network.

Context.—: The Oncotype DX Recurrence Score (RS) predicts recurrence and chemotherapy benefit in early-stage estrogen receptor-positive breast cancer patients. Cost and unavailability are 2 major disadvantages of the assay. Multiple models have been developed to predict the RS., Objective.—: To predict RS based on histopathologic and biomarker features, and to measure concordance and correlation with RS of the following 3 algorithms: breast cancer prognostic score, Magee0, and Magee2., Design.—: Breast cancer cases with available RSs were reviewed (n = 442). RS categories were stratified by pathologic and biomarker variables. Histopathologic and biomarker data were abstracted from pathology reports, and RS was calculated by each model. Correlation and concordance between models and RS were calculated., Results.—: Less than 5% of breast cancers with lobular features, low-grade tumors, carcinomas with high progesterone receptor content, or luminal A tumors had an RS greater than 25. Breast cancer prognostic score, Magee0, and Magee2 demonstrated correlation coefficients with RS of 0.63, 0.61, and 0.62, respectively. Two-step discordances were uncommon. When an RS of 25 was used to separate high-risk from non-high-risk cases, concordance rates of 86% to 88% were achieved., Conclusions.—: High RS was observed only in a small percentage of pure or mixed lobular carcinomas, low-grade or luminal A tumors, and tumors with high progesterone receptor expression, suggesting that these cancers may not require Oncotype testing. All 3 surrogate models demonstrated comparable correlation and high concordance with the RS when a cutoff of 25 was used, suggesting their utility in cases where the actual RS is unavailable.

Background: Cell-free DNA noninvasive prenatal screening for trisomies 21, 18, and 13 has been rapidly adopted into clinical practice. However, previous studies are limited by a lack of follow-up genetic testing to confirm the outcomes and accurately assess test performance, particularly in women at a low risk for aneuploidy., Objective: To measure and compare the performance of cell-free DNA screening for trisomies 21, 18, and 13 between women at a low and high risk for aneuploidy in a large, prospective cohort with genetic confirmation of results STUDY DESIGN: This was a multicenter prospective observational study at 21 centers in 6 countries. Women who had single-nucleotide-polymorphism-based cell-free DNA screening for trisomies 21, 18, and 13 were enrolled. Genetic confirmation was obtained from prenatal or newborn DNA samples. The test performance and test failure (no-call) rates were assessed for the cohort, and women with low and high previous risks for aneuploidy were compared. An updated cell-free DNA algorithm blinded to the pregnancy outcome was also assessed., Results: A total of 20,194 women were enrolled at a median gestational age of 12.6 weeks (interquartile range, 11.6-13.9). The genetic outcomes were confirmed in 17,851 cases (88.4%): 13,043 (73.1%) low-risk and 4808 (26.9%) high-risk cases for aneuploidy. Overall, 133 trisomies were diagnosed (100 trisomy 21; 18 trisomy 18; 15 trisomy 13). The cell-free DNA screen positive rate was lower in the low-risk vs the high-risk group (0.27% vs 2.2%; P<.0001). The sensitivity and specificity were similar between the groups. The positive predictive value for the low- and high-risk groups was 85.7% vs 97.5%; P=.058 for trisomy 21; 50.0% vs 81.3%; P=.283 for trisomy 18; and 62.5% vs 83.3; P=.58 for trisomy 13, respectively. Overall, 602 (3.4%) patients had no-call result after the first draw and 287 (1.61%) after including cases with a second draw. The trisomy rate was higher in the 287 cases with no-call results than patients with a result on a first draw (2.8% vs 0.7%; P=.001). The updated algorithm showed similar sensitivity and specificity to the study algorithm with a lower no-call rate., Conclusion: In women at a low risk for aneuploidy, single-nucleotide-polymorphism-based cell-free DNA has high sensitivity and specificity, positive predictive value of 85.7% for trisomy 21 and 74.3% for the 3 common trisomies. Patients who receive a no-call result are at an increased risk of aneuploidy and require additional investigation., (Copyright © 2022 The Author(s). Published by Elsevier Inc. All rights reserved.)

Background: Persons experiencing homelessness (PEH) often use hospital Emergency Department (ED) as the only source of healthcare. The aim of this study was to undertake a systematic review to identify the prevalence, clinical reasons and outcomes in relation to ED visits by PEH., Methods: A protocol-led (CRD42020189263) systematic review was conducted using search of MEDLINE, EMBASE, CINAHL and Google Scholar databases. Studies that reported either the prevalence of homelessness in the ED or clinical reasons for presentation to ED by PEH and published in English language were included. Definitions of homelessness used by study authors were accepted., Results: From the screening of 1349 unique titles, a total of 36 studies were included. Wide variations in the prevalence and key cause of presentations were identified across the studies often linked to differences in country, study setting, disease classification and data collection methods. The proportion of ED visits contributed by PEH ranged from 0.41 to 19.6%. PEH made an average of 0.72 visits to 5.8 visits per person per year in the ED [rate ratio compared to non-homeless 1.63 to 18.75]. Up to a third and quarter of the visits were contributed by alcohol-related diagnoses and substance poisoning respectively. The percentage of PEH who died in the ED ranged from 0.1 to 0.5%., Conclusions: Drug-, alcohol- and injury-related presentations dominate the ED visits by PEH. Wide variations in the data were observed in regard to attendance and treatment outcomes. There is a need for prevention actions in the community, integrated discharge and referral pathways between health, housing and social care to minimise frequent usage and improve attendance outcomes., (© 2022. The Author(s).)

Women of African ancestry suffer higher rates of breast cancer mortality compared with all other groups in the United States. Though the precise reasons for these disparities remain unclear, many recent studies have implicated a role for differences in tumor biology. Using an epitope-validated antibody against the endoplasmic reticulum-associated E3 ligase, gp78, we show that elevated levels of gp78 in patient breast cancer cells predict poor survival. Moreover, high levels of gp78 are associated with poor outcomes in both ER+ and ER- tumors, and breast cancers expressing elevated amounts of gp78 protein are enriched in gene expression pathways that influence cell cycle, metabolism, receptor-mediated signaling, and cell stress response pathways. In multivariate analysis adjusted for subtype and grade, gp78 protein is an independent predictor of poor outcomes in women of African ancestry. Furthermore, gene expression signatures, derived from patients stratified by gp78 protein expression, are strong predictors of recurrence and pathological complete response in retrospective clinical trial data and share many common features with gene sets previously identified to be overrepresented in breast cancers based on race. These findings implicate a prominent role for gp78 in tumor progression and offer insights into our understanding of racial differences in breast cancer outcomes.

Background: Historically, prenatal screening has focused primarily on the detection of fetal aneuploidies. Cell-free DNA now enables noninvasive screening for subchromosomal copy number variants, including 22q11.2 deletion syndrome (or DiGeorge syndrome), which is the most common microdeletion and a leading cause of congenital heart defects and neurodevelopmental delay. Although smaller studies have demonstrated the feasibility of screening for 22q11.2 deletion syndrome, large cohort studies with confirmatory postnatal testing to assess test performance have not been reported., Objective: This study aimed to assess the performance of single-nucleotide polymorphism-based, prenatal cell-free DNA screening for detection of 22q11.2 deletion syndrome., Study Design: Patients who underwent single-nucleotide polymorphism-based prenatal cell-free DNA screening for 22q11.2 deletion syndrome were prospectively enrolled at 21 centers in 6 countries. Prenatal or newborn DNA samples were requested in all cases for genetic confirmation using chromosomal microarrays. The primary outcome was sensitivity, specificity, positive predictive value, and negative predictive value of cell-free DNA screening for the detection of all deletions, including the classical deletion and nested deletions that are ≥500 kb, in the 22q11.2 low-copy repeat A-D region. Secondary outcomes included the prevalence of 22q11.2 deletion syndrome and performance of an updated cell-free DNA algorithm that was evaluated with blinding to the pregnancy outcome., Results: Of the 20,887 women enrolled, a genetic outcome was available for 18,289 (87.6%). A total of 12 22q11.2 deletion syndrome cases were confirmed in the cohort, including 5 (41.7%) nested deletions, yielding a prevalence of 1 in 1524. In the total cohort, cell-free DNA screening identified 17,976 (98.3%) cases as low risk for 22q11.2 deletion syndrome and 38 (0.2%) cases as high risk; 275 (1.5%) cases were nonreportable. Overall, 9 of 12 cases of 22q11.2 were detected, yielding a sensitivity of 75.0% (95% confidence interval, 42.8-94.5); specificity of 99.84% (95% confidence interval, 99.77-99.89); positive predictive value of 23.7% (95% confidence interval, 11.44-40.24), and negative predictive value of 99.98% (95% confidence interval, 99.95-100). None of the cases with a nonreportable result was diagnosed with 22q11.2 deletion syndrome. The updated algorithm detected 10 of 12 cases (83.3%; 95% confidence interval, 51.6-97.9) with a lower false positive rate (0.05% vs 0.16%; P<.001) and a positive predictive value of 52.6% (10/19; 95% confidence interval, 28.9-75.6)., Conclusion: Noninvasive cell-free DNA prenatal screening for 22q11.2 deletion syndrome can detect most affected cases, including smaller nested deletions, with a low false positive rate., (Copyright © 2022 The Author(s). Published by Elsevier Inc. All rights reserved.)

Introduction: Although minimally invasive surgery (MIS) has clearly been associated with improved colorectal surgery outcomes, not all populations benefit from this approach. Using a national database, we analyzed both, the trend in the utilization of MIS for diverticulitis and differences in utilization by race., Methods: Colon-targeted participant user files (PUFs) from 2012 to 18 were linked to respective PUFs in National Surgical Quality Improvement Project. Patients undergoing colectomy for acute diverticulitis or chronic diverticular disease were included. Surgical approach was stratified by race and year. To adjust for confounding and estimate the association of covariates with approach, data were fit using multivariable binary logistic regression main effects model. Using a joint effects model, we evaluated whether the odds of a particular approach over time was differentially affected by race., Results: Of the 46 713 patients meeting inclusion criteria, 83% were white, with 7% black and 10% other. Over the study period, there was a decrease in the rate of open colectomy of about 5% P < .001, and increase in the rate of utilization of laparoscopic and robotic approaches (RC) P < .0001. After adjusting for confounders, black race was associated with open surgery P < .0001., Conclusion: There is disparity in the utilization of MIS for diverticulitis. Further research into the reasons for this disparity is critical to ensure known benefits of MIC are realized across all races.

Terahertz imaging and spectroscopy is an exciting technology that has the potential to provide insights in medical imaging. Prior research has leveraged statistical inference to classify tissue regions from terahertz images. To date, these approaches have shown that the segmentation problem is challenging for images of fresh tissue and for tumors that have invaded muscular regions. Artificial intelligence, particularly machine learning and deep learning, has been shown to improve performance in some medical imaging challenges. This paper builds on that literature by modifying a set of deep learning approaches to the challenge of classifying tissue regions of images captured by terahertz imaging and spectroscopy of freshly excised murine xenograft tissue. Our approach is to preprocess the images through a wavelet synchronous-squeezed transformation (WSST) to convert time-sequential terahertz data of each THz pixel to a spectrogram. Spectrograms are used as input tensors to a deep convolution neural network for pixel-wise classification. Based on the classification result of each pixel, a cancer tissue segmentation map is achieved. In experimentation, we adopt leave-one-sample-out cross-validation strategy, and evaluate our chosen networks and results using multiple metrics such as accuracy, precision, intersection, and size. The results from this experimentation demonstrate improvement in classification accuracy compared to statistical methods, an improvement to segmentation between muscle and cancerous regions in xenograft tumors, and identify areas to improve the imaging and classification methodology., Competing Interests: Conflict of Interest The authors declare no competing interests.

Purpose: We investigate the enhancement in terahertz (THz) images of freshly excised breast tumors upon treatment with an optical clearance agent. The hyperspectral imaging and spectral classifications are used to quantitatively demonstrate the image enhancement. Glycerol solution with 60% concentration is applied to excised breast tumor specimens for various time durations to investigate the effectiveness on image enhancement. Approach: THz reflection spectroscopy is utilized to obtain the absorption coefficient and the index of refraction of untreated and glycerol-treated tissues at each frequency up to 3 THz. Two classifiers, spectral angular mapping (SAM) based on several kernels and Euclidean minimum distance (EMD) are implemented to evaluate the effectiveness of the treatment. The testing raw data is obtained from five breast cancer specimens: two untreated specimens and three specimens treated with glycerol solution for 20, 40, or 60 min. All tumors used in the testing data have healthy tissues adjacent to cancerous ones consistent with the challenge faced in lumpectomy surgeries. Results: The glycerol-treated tissues showed a decrease in the absorption coefficients compared with untreated tissues, especially as the period of treatment increased. Although the sensitivity metric of the classifier presented higher values in the untreated tissues compared with the treated ones, the specificity and accuracy metrics demonstrated higher values for the treated tissues compared with the untreated ones. Conclusions: The biocompatible glycerol solution is a potential optical clearance agent in THz imaging while keeping the histopathology imaging intact. The SAM technique provided a good classification of cancerous tissues despite the small amount of cancer in the training data (only 7%). The SAM exponential kernel and EMD presented classification accuracy of ∼ 80 % to 85% compared with linear and polynomial kernels that provided accuracy ranging from 70% to 80%. Overall, glycerol treatment provides a potential improvement in cancer classification in freshly excised breast tumors., (© 2022 Society of Photo-Optical Instrumentation Engineers (SPIE).)