Tool use increases mechanical foraging success and tooth health in southern sea otters (Enhydra lutris nereis).

Although tool use may enhance resource utilization, its fitness benefits are difficult to measure. By examining longitudinal data from 196 radio-tagged southern sea otters (Enhydra lutris nereis), we found that tool-using individuals, particularly females, gained access to larger and/or harder-shelled prey. These mechanical advantages translated to reduced tooth damage during food processing. We also found that tool use diminishes trade-offs between access to different prey, tooth condition, and energy intake, all of which are dependent on the relative prey availability in the environment. Tool use allowed individuals to maintain energetic requirements through the processing of alternative prey that are typically inaccessible with biting alone, suggesting that this behavior is a necessity for the survival of some otters in environments where preferred prey are depleted.

Disparities in Receipt of National Comprehensive Cancer Network Guideline-Adherent Care and Outcomes among Women with Triple-Negative Breast Cancer by Race/Ethnicity, Socioeconomic Status, and Insurance Type.

BACKGROUND: The National Comprehensive Cancer Network guidelines were designed to improve patient outcomes. Here, we examine factors that may contribute to outcomes and guideline adherence in patients with triple-negative breast cancer. METHODS: This was a retrospective cohort study of women with triple-negative breast cancer using the California Cancer Registry. Adherent treatment was defined as the receipt of a combination of surgery, lymph node assessment, adjuvant radiation, and/or chemotherapy. A multivariable logistic regression was used to determine the effects of independent variables on adherence to the NCCN guidelines. Disease-specific survival was calculated using Cox regression analysis. RESULTS: A total of 16,858 women were analyzed. Black and Hispanic patients were less likely to receive guideline-adherent care (OR 0.82, 95%CI 0.73-0.92 and OR 0.87, 95%CI 0.79-0.95, respectively) compared to White patients. Hazard ratios adjusted for adherent care showed that Black patients had increased disease-specific mortality (HR 1.28, 95%CI 1.16-1.42, p < 0.0001) compared to White patients. CONCLUSIONS: A significant majority of breast cancer patients in California continue to receive non-guideline-adherent care. Non-Hispanic Black patients and patients from lower SES quintile groups were less likely to receive guideline-adherent care. Patients with non-adherent care had worse disease-specific survival compared to recipients of NCCN guideline-adherent care.

Effects of CYP3A4 and CYP2C9 genotype on systemic anastrozole and fulvestrant concentrations in SWOG S0226.

Objective & methods: This study tested associations of genotype-predicted activity of CYP3A4, other pharmacogenes, SLC28A7 (rs11648166) and ALPPL2 (rs28845026) with systemic concentrations of the endocrine therapies anastrozole and fulvestrant in SWOG S0226 trial participants. Results: Participants in the anastrozole-only arm with low CYP3A4 activity (i.e. CYP3A4*22 carriers) had higher systemic anastrozole concentrations than patients with high CYP3A4 activity (ß-coefficient = 10.03; 95% CI: 1.42, 18.6; p = 0.025). In an exploratory analysis, participants with low CYP2C9 activity had lower anastrozole concentrations and higher fulvestrant concentrations than participants with high CYP2C9 activity. Conclusion: Inherited genetic variation in CYP3A4 and CYP2C9 may affect concentrations of endocrine therapy and may be useful to personalize dosing and improve treatment outcomes.

Thymidine kinase activity levels in serum can identify HR+ metastatic breast cancer patients with a low risk of early progression (SWOG S0226).

BACKGROUND: Some patients with metastatic breast cancer (MBC) stay on endocrine therapy (ET) for years and others progress quickly. Serum thymidine kinase activity (TKa), an indicator of cell-proliferation, is a potential biomarker for monitoring ET and predicting MBC outcome. We have previously reported TKa as being prognostic in MBC in SWOG S0226. Here, new data on progression within 30/60 days post sampling, with a new, FDA approved version of DiviTum®TKa highlighting differences vs. a Research Use Only version is reported. METHODS: 1,546 serum samples from 454 patients were assessed, collected at baseline and at 4 subsequent timepoints during treatment. A new predefined cut-off tested the ability to predict disease progression. A new measuring unit, DuA (DiviTum® unit of Activity) is adopted. RESULTS: A DiviTum®TKa score <250 DuA provides a much lower risk of progression within 30/60 days after blood draw, the negative predictive value (NPV) was 96.7% and 93.5%, respectively. Patients <250 DuA experienced significantly longer progression-free survival and overall survival, demonstrated at baseline and for all time intervals. CONCLUSIONS: DiviTum®TKa provides clinically meaningful information for patients with HR+ MBC. Low TKa levels provide such a high NPV for rapid progression that such patients might forego additional therapy added to single agent ET.Trial registration: NCT00075764.

Deep Learning-based Automatic Diagnosis of Breast Cancer on MRI Using Mask R-CNN for Detection Followed by ResNet50 for Classification.

RATIONALE AND OBJECTIVES: Diagnosis of breast cancer on MRI requires, first, the identification of suspicious lesions; second, the characterization to give a diagnostic impression. We implemented Mask Reginal-Convolutional Neural Network (R-CNN) to detect abnormal lesions, followed by ResNet50 to estimate the malignancy probability. MATERIALS AND METHODS: Two datasets were used. The first set had 176 cases, 103 cancer, and 73 benign. The second set had 84 cases, 53 cancer, and 31 benign. For detection, the pre-contrast image and the subtraction images of left and right breasts were used as inputs, so the symmetry could be considered. The detected suspicious area was characterized by ResNet50, using three DCE parametric maps as inputs. The results obtained using slice-based analyses were combined to give a lesion-based diagnosis. RESULTS: In the first dataset, 101 of 103 cancers were detected by Mask R-CNN as suspicious, and 99 of 101 were correctly classified by ResNet50 as cancer, with a sensitivity of 99/103 = 96%. 48 of 73 benign lesions and 131 normal areas were identified as suspicious. Following classification by ResNet50, only 16 benign and 16 normal areas remained as malignant. The second dataset was used for independent testing. The sensitivity was 43/53 = 81%. Of the total of 121 identified non-cancerous lesions, only 6 of 31 benign lesions and 22 normal tissues were classified as malignant. CONCLUSION: ResNet50 could eliminate approximately 80% of false positives detected by Mask R-CNN. Combining Mask R-CNN and ResNet50 has the potential to develop a fully-automatic computer-aided diagnostic system for breast cancer on MRI.

Effects of acute temperature change on California moray prey manipulation and transport behavior.

California moray eels, Gymnothorax mordax, are benthic predatory residents of southern California kelp forest ecosystems. California morays around Catalina Island move vertically through the water column to feed, exposing them to a wide range of temperatures. For a predatory fish, morays have a relatively large prey handling repertoire that enable them to manipulate their prey before swallowing. Prey manipulation behaviors include shaking, spinning, knotting, and ramming prey against other objects. Morays also have observable transport mechanics where they protract and retract their pharyngeal jaws to swallow prey. We examined prey manipulation and transport behaviors at four temperature treatments that simulated the range of environmental temperatures morays encounter in the wild. We hypothesized that higher temperatures will increase the prevalence, duration, and rate of whole body prey manipulation behaviors and decrease the duration of prey transport time. Previous temperature studies focused on fishes occupying intermediate trophic levels. Therefore, understanding how acute temperature affects feeding behavior of the California moray eel, an abundant predatory fish, is especially important, as changes in environmental temperature may have disproportionate effects in their marine community. Five morays were acutely exposed to 15, 18, 21, 24 °C temperatures and their subsequent feeding behaviors were filmed and quantified. Individuals were offered the same relative prey mass (15 %) in relation to their body mass throughout the study. We compared the number of times each prey manipulation behavior occurred, the mean time morays employed each behavior, and the rate (number of times per second) each behavior was performed across different temperatures. Our data demonstrates that absolute time spent knotting varies significantly across temperature. Knotting, often used to remove pieces from larger prey, was most frequent at 21 and 24 °C. The average duration of knotting also increased with temperature. The rates of prey manipulation behaviors did not vary significantly with temperature. Finally, transport behavior did not vary across treatments. Our study shows that knotting behavior in the California moray is responsive to environmental temperatures and that morays may be able to manipulate larger prey in warmer waters. These behavioral data may have important implications for predator-prey relationships under dynamic and future ocean conditions.

Trade-offs between foraging reward and mortality risk drive sex-specific foraging strategies in sexually dimorphic northern elephant seals.

Sex-specific phenotypic differences are widespread throughout the animal kingdom. Reproductive advantages provided by trait differences come at a cost. Here, we link sex-specific foraging strategies to trade-offs between foraging reward and mortality risk in sexually dimorphic northern elephant seals (Mirounga angustirostris). We analyse a decadal dataset on movement patterns, dive behaviour, foraging success and mortality rates. Females are deep-diving predators in open ocean habitats. Males are shallow-diving benthic predators in continental shelf habitats. Males gain six times more mass and acquire energy 4.1 times faster than females. High foraging success comes with a high mortality rate. Males are six times more likely to die than females. These foraging strategies and trade-offs are related to different energy demands and life-history strategies. Males use a foraging strategy with a high mortality risk to attain large body sizes necessary to compete for females, as only a fraction of the largest males ever mate. Females use a foraging strategy with a lower mortality risk, maximizing reproductive success by pupping annually over a long lifespan. Our results highlight how sex-specific traits can drive disparity in mortality rates and expand species' niche space. Further, trade-offs between foraging rewards and mortality risk can differentially affect each sex's ability to maximize fitness.

Automatic Detection and Segmentation of Breast Cancer on MRI Using Mask R-CNN Trained on Non-Fat-Sat Images and Tested on Fat-Sat Images.

RATIONALE AND OBJECTIVES: Computer-aided methods have been widely applied to diagnose lesions on breast magnetic resonance imaging (MRI). The first step was to identify abnormal areas. A deep learning Mask Regional Convolutional Neural Network (R-CNN) was implemented to search the entire set of images and detect suspicious lesions. MATERIALS AND METHODS: Two DCE-MRI datasets were used, 241 patients acquired using non-fat-sat sequence for training, and 98 patients acquired using fat-sat sequence for testing. All patients have confirmed unilateral mass cancers. The tumor was segmented using fuzzy c-means clustering algorithm to serve as the ground truth. Mask R-CNN was implemented with ResNet-101 as the backbone. The neural network output the bounding boxes and the segmented tumor for evaluation using the Dice Similarity Coefficient (DSC). The detection performance, and the trade-off between sensitivity and specificity, was analyzed using free response receiver operating characteristic. RESULTS: When the precontrast and subtraction image of both breasts were used as input, the false positive from the heart and normal parenchymal enhancements could be minimized. The training set had 1469 positive slices (containing lesion) and 9135 negative slices. In 10-fold cross-validation, the mean accuracy = 0.86 and DSC = 0.82. The testing dataset had 1568 positive and 7264 negative slices, with accuracy = 0.75 and DSC = 0.79. When the obtained per-slice results were combined, 240 of 241 (99.5%) lesions in the training and 98 of 98 (100%) lesions in the testing datasets were identified. CONCLUSION: Deep learning using Mask R-CNN provided a feasible method to search breast MRI, localize, and segment lesions. This may be integrated with other artificial intelligence algorithms to develop a fully automatic breast MRI diagnostic system.

Diagnosis of Breast Cancer Using Radiomics Models Built Based on Dynamic Contrast Enhanced MRI Combined With Mammography.

OBJECTIVE: To build radiomics models using features extracted from DCE-MRI and mammography for diagnosis of breast cancer. MATERIALS AND METHODS: 266 patients receiving MRI and mammography, who had well-enhanced lesions on MRI and histologically confirmed diagnosis were analyzed. Training dataset had 146 malignant and 56 benign, and testing dataset had 48 malignant and 18 benign lesions. Fuzzy-C-means clustering algorithm was used to segment the enhanced lesion on subtraction MRI maps. Two radiologists manually outlined the corresponding lesion on mammography by consensus, with the guidance of MRI maximum intensity projection. Features were extracted using PyRadiomics from three DCE-MRI parametric maps, and from the lesion and a 2-cm bandshell margin on mammography. The support vector machine (SVM) was applied for feature selection and model building, using 5 datasets: DCE-MRI, mammography lesion-ROI, mammography margin-ROI, mammography lesion+margin, and all combined. RESULTS: In the training dataset evaluated using 10-fold cross-validation, the diagnostic accuracy of the individual model was 83.2% for DCE-MRI, 75.7% for mammography lesion, 64.4% for mammography margin, and 77.2% for lesion+margin. When all features were combined, the accuracy was improved to 89.6%. By adding mammography features to MRI, the specificity was significantly improved from 69.6% (39/56) to 82.1% (46/56), p<0.01. When the developed models were applied to the independent testing dataset, the accuracy was 78.8% for DCE-MRI and 83.3% for combined MRI+Mammography. CONCLUSION: The radiomics model built from the combined MRI and mammography has the potential to provide a machine learning-based diagnostic tool and decrease the false positive diagnosis of contrast-enhanced benign lesions on MRI.

BI-RADS Reading of Non-Mass Lesions on DCE-MRI and Differential Diagnosis Performed by Radiomics and Deep Learning.

BACKGROUND: A wide variety of benign and malignant processes can manifest as non-mass enhancement (NME) in breast MRI. Compared to mass lesions, there are no distinct features that can be used for differential diagnosis. The purpose is to use the BI-RADS descriptors and models developed using radiomics and deep learning to distinguish benign from malignant NME lesions. MATERIALS AND METHODS: A total of 150 patients with 104 malignant and 46 benign NME were analyzed. Three radiologists performed reading for morphological distribution and internal enhancement using the 5th BI-RADS lexicon. For each case, the 3D tumor mask was generated using Fuzzy-C-Means segmentation. Three DCE parametric maps related to wash-in, maximum, and wash-out were generated, and PyRadiomics was applied to extract features. The radiomics model was built using five machine learning algorithms. ResNet50 was implemented using three parametric maps as input. Approximately 70% of earlier cases were used for training, and 30% of later cases were held out for testing. RESULTS: The diagnostic BI-RADS in the original MRI report showed that 104/104 malignant and 36/46 benign lesions had a BI-RADS score of 4A-5. For category reading, the kappa coefficient was 0.83 for morphological distribution (excellent) and 0.52 for internal enhancement (moderate). Segmental and Regional distribution were the most prominent for the malignant group, and focal distribution for the benign group. Eight radiomics features were selected by support vector machine (SVM). Among the five machine learning algorithms, SVM yielded the highest accuracy of 80.4% in training and 77.5% in testing datasets. ResNet50 had a better diagnostic performance, 91.5% in training and 83.3% in testing datasets. CONCLUSION: Diagnosis of NME was challenging, and the BI-RADS scores and descriptors showed a substantial overlap. Radiomics and deep learning may provide a useful CAD tool to aid in diagnosis.

Evaluating Serum Thymidine Kinase 1 in Patients with Hormone Receptor-Positive Metastatic Breast Cancer Receiving First-line Endocrine Therapy in the SWOG S0226 Trial.

PURPOSE: Serum thymidine kinase 1 (sTK1) activity is associated with poor prognosis in metastatic breast cancer (MBC). We assessed the prognostic effect of sTK1 in patients with hormone receptor-positive MBC treated on a prospective randomized trial of anastrozole (A) versus A plus fulvestrant (A + F). PATIENTS AND METHODS: sTK1 was assessed in 1,726 serums [baseline (BL), cycles 2, 3, 4, and 7] using the DiviTum assay. A prespecified cutoff of ≥200 Du/L was considered high. Progression-free survival (PFS) and overall survival (OS) were analyzed by Kaplan-Meier, log-rank tests, and Cox regression. RESULTS: BL sTK1 was elevated in 171 (40%) of 432 patients. Patients with high versus low BL sTK1 had significantly worse PFS [median 11.2 vs. 17.3 months, HR = 1.76; 95% confidence interval (CI; 1.43-2.16); P < 0.0001] and OS [median 30 vs. 58 months, HR = 2.38; 95% CI (1.91-2.98); P < 0.0001]. OS was significantly better for patients with high sTK1 who did not have prior adjuvant tamoxifen and who received A + F versus A alone [median 46 vs. 21 months, HR = 0.58; 95% CI (0.38-0.87); P = 0.0087]. Patients with low sTK1 had no difference in outcomes by therapy (P = 0.44). At serial timepoints, high versus low sTK1 had significantly worse subsequent PFS and OS [at cycle 2: PFS HR = 1.70, 95% CI (1.34-2.17); P < 0.0001, OS HR = 2.51, 95% CI (1.93-3.26); P < 0.0001]. CONCLUSIONS: High sTK1 at BL and subsequent timepoints is associated with worse prognosis in patients with MBC starting first-line endocrine therapy (ET). Patients with low sTK1 at BL have comparable outcomes on single-agent or combination ET.

Anti-racist interventions to transform ecology, evolution and conservation biology departments.

Racial and ethnic discrimination persist in science, technology, engineering and mathematics fields, including ecology, evolution and conservation biology (EECB) and related disciplines. Marginalization and oppression as a result of institutional and structural racism continue to create barriers to inclusion for Black people, Indigenous people and people of colour (BIPOC), and remnants of historic racist policies and pseudoscientific theories continue to plague these fields. Many academic EECB departments seek concrete ways to improve the climate and implement anti-racist policies in their teaching, training and research activities. We present a toolkit of evidence-based interventions for academic EECB departments to foster anti-racism in three areas: in the classroom; within research laboratories; and department wide. To spark restorative discussion and action in these areas, we summarize EECB's racist and ethnocentric histories, as well as current systemic problems that marginalize non-white groups. Finally, we present ways that EECB departments can collectively address shortcomings in equity and inclusion by implementing anti-racism, and provide a positive model for other departments and disciplines.

Asymmetry in the frequency and proportion of arm truncation in three sympatric California Octopus species.

Octopuses have eight radially symmetrical arms that surround the base of a bilaterally symmetrical body. These numerous appendages, which explore the environment, handle food, and defend the animal against predators, are highly susceptible to truncation or loss. Here, we used scaling relationships specific to the arms of three sympatric octopus species of the genus Octopus, to calculate the proportion of arm truncation. We then compared the frequency and proportion of arm losses between different body locations. Truncated arms were found in 59.8 % of specimens examined, with individuals bearing one to as many as seven injured arms. We found a significant left side bias for greater proportion of arm truncation for all species and sexes except in O. bimaculatus males. We also found that sister species O. bimaculatus and O. bimaculoides had a greater proportion of their anterior arms (pairs 1 and 2) truncated, while in O. rubescens, posterior arms (pairs 3 and 4) were more truncated. The mean percent of arm that was truncated was 28.1 % overall but varied between species and by sex and was highest in O. rubescens females (56 %). The arms of O. rubescens also exhibited the steepest scaling patterns, and showed a positive correlation between body size and number of truncated arms. Overall, we show that arm injuries in our sampling of three intertidal species are frequent and asymmetrical, and that when injured, octopus on average lose a considerable proportion of their arm. Through quantifying the variation in arm truncation, this study provides a new foundation to explore behavioral compensation for arm loss in cephalopods.

Development of U-Net Breast Density Segmentation Method for Fat-Sat MR Images Using Transfer Learning Based on Non-Fat-Sat Model.

To develop a U-net deep learning method for breast tissue segmentation on fat-sat T1-weighted (T1W) MRI using transfer learning (TL) from a model developed for non-fat-sat images. The training dataset (N = 126) was imaged on a 1.5 T MR scanner, and the independent testing dataset (N = 40) was imaged on a 3 T scanner, both using fat-sat T1W pulse sequence. Pre-contrast images acquired in the dynamic-contrast-enhanced (DCE) MRI sequence were used for analysis. All patients had unilateral cancer, and the segmentation was performed using the contralateral normal breast. The ground truth of breast and fibroglandular tissue (FGT) segmentation was generated using a template-based segmentation method with a clustering algorithm. The deep learning segmentation was performed using U-net models trained with and without TL, by using initial values of trainable parameters taken from the previous model for non-fat-sat images. The ground truth of each case was used to evaluate the segmentation performance of the U-net models by calculating the dice similarity coefficient (DSC) and the overall accuracy based on all pixels. Pearson's correlation was used to evaluate the correlation of breast volume and FGT volume between the U-net prediction output and the ground truth. In the training dataset, the evaluation was performed using tenfold cross-validation, and the mean DSC with and without TL was 0.97 vs. 0.95 for breast and 0.86 vs. 0.80 for FGT. When the final model developed with and without TL from the training dataset was applied to the testing dataset, the mean DSC was 0.89 vs. 0.83 for breast and 0.81 vs. 0.81 for FGT, respectively. Application of TL not only improved the DSC, but also decreased the required training case number. Lastly, there was a high correlation (R2 > 0.90) for both the training and testing datasets between the U-net prediction output and ground truth for breast volume and FGT volume. U-net can be applied to perform breast tissue segmentation on fat-sat images, and TL is an efficient strategy to develop a specific model for each different dataset.

Snowflake morays, Echidna nebulosa, exhibit similar feeding kinematics in terrestrial and aquatic treatments.

Some species of durophagous moray eels (Muraenidae) have been documented emerging from the marine environment to capture intertidal crabs but how they consume prey out of water is unknown. Here, we trained snowflake morays, Echidna nebulosa, to undulate out of the aquatic environment to feed on land. On land, snowflake morays remove prey from the substrate by biting and swallow prey using pharyngeal jaw enabled transport. Although snowflake morays exhibit smaller jaw rotation angles on land when apprehending their prey, transport kinematics involving dorsoventral flexion of the head to protract the pharyngeal jaws and overall feeding times did not differ between terrestrial and aquatic treatments. We suggest that their elongate body plan, ability to rotate their heads in the dorsoventral and lateral directions, and extreme pharyngeal movements all contribute to the ability of durophagous morays to feed in the terrestrial environment.

Body shape transformations by alternate anatomical adaptive peak shifts in blenniiform fishes.

Extreme body elongation has occurred repeatedly in the evolutionary history of ray-finned fishes. Lengthening of the anterior-posterior body axis relative to depth and width can involve changes in the cranial skeleton and vertebral column, but to what extent is anatomical evolution determined by selective factors and intrinsic constraints that are shared broadly among closely related lineages? In this study, we fit adaptive (Ornstein-Uhlenbeck) evolutionary models to body shape and its anatomical determinants and identified two instances of extreme elongation by divergent anatomical peak shifts in the Blenniiformes, a radiation of small-bodied substrate-associated marine teleost fishes. Species in the genus Xiphasia (hairtail blennies) evolved toward a peak defined by a highly elongated caudal vertebral region but ancestral cranial and precaudal vertebral morphology. In contrast, a clade that includes the genera Chaenopsis and Lucayablennius (pike and arrow blennies) evolved toward a peak with a long slender skull but ancestral axial skeletal anatomy. Neither set of anatomical peak shifts aligns closely with the major axis of anatomical diversification in other blenniiform fishes. These results provide little evidence that ancestral constraints have affected body shape transformation, and instead suggest that extreme elongation arose with distinct shifts in selective factors and development.

Breast cancer differential diagnosis using diffuse optical spectroscopic imaging and regression with z-score normalized data.

SIGNIFICANCE: Current imaging paradigms for differential diagnosis of suspicious breast lesions suffer from high false positive rates that force patients to undergo unnecessary biopsies. Diffuse optical spectroscopic imaging (DOSI) noninvasively probes functional hemodynamic and compositional parameters in deep tissue and has been shown to be sensitive to contrast between normal and malignant tissues. AIM: DOSI methods are under investigation as an adjunct to mammography and ultrasound that could reduce false positive rates and unnecessary biopsies, particularly in radiographically dense breasts. METHODS: We performed a retrospective analysis of 212 subjects with suspicious breast lesions who underwent DOSI imaging. Physiological tissue parameters were z-score normalized to the patient's contralateral breast tissue and input to univariate logistic regression models to discriminate between malignant tumors and the surrounding normal tissue. The models were then used to differentiate malignant lesions from benign lesions. RESULTS: Models incorporating several individual hemodynamic parameters were able to accurately distinguish malignant tumors from both the surrounding background tissue and benign lesions with area under the curve (AUC) ≥0.85. Z-score normalization improved the discriminatory ability and calibration of these predictive models relative to unnormalized or ratio-normalized data. CONCLUSIONS: Findings from a large subject population study show how DOSI data normalization that accounts for normal tissue heterogeneity and quantitative statistical regression approaches can be combined to improve the ability of DOSI to diagnose malignant lesions. This improved diagnostic accuracy, combined with the modality's inherent logistical advantages of portability, low cost, and nonionizing radiation, could position DOSI as an effective adjunct modality that could be used to reduce the number of unnecessary invasive biopsies.

Prediction of breast cancer molecular subtypes on DCE-MRI using convolutional neural network with transfer learning between two centers.

OBJECTIVES: To apply deep learning algorithms using a conventional convolutional neural network (CNN) and a recurrent CNN to differentiate three breast cancer molecular subtypes on MRI. METHODS: A total of 244 patients were analyzed, 99 in training dataset scanned at 1.5 T and 83 in testing-1 and 62 in testing-2 scanned at 3 T. Patients were classified into 3 subtypes based on hormonal receptor (HR) and HER2 receptor: (HR+/HER2-), HER2+, and triple negative (TN). Only images acquired in the DCE sequence were used in the analysis. The smallest bounding box covering tumor ROI was used as the input for deep learning to develop the model in the training dataset, by using a conventional CNN and the convolutional long short-term memory (CLSTM). Then, transfer learning was applied to re-tune the model using testing-1(2) and evaluated in testing-2(1). RESULTS: In the training dataset, the mean accuracy evaluated using tenfold cross-validation was higher by using CLSTM (0.91) than by using CNN (0.79). When the developed model was applied to the independent testing datasets, the accuracy was 0.4-0.5. With transfer learning by re-tuning parameters in testing-1, the mean accuracy reached 0.91 by CNN and 0.83 by CLSTM, and improved accuracy in testing-2 from 0.47 to 0.78 by CNN and from 0.39 to 0.74 by CLSTM. Overall, transfer learning could improve the classification accuracy by greater than 30%. CONCLUSIONS: The recurrent network using CLSTM could track changes in signal intensity during DCE acquisition, and achieved a higher accuracy compared with conventional CNN during training. For datasets acquired using different settings, transfer learning can be applied to re-tune the model and improve accuracy. KEY POINTS: • Deep learning can be applied to differentiate breast cancer molecular subtypes. • The recurrent neural network using CLSTM could track the change of signal intensity in DCE images, and achieved a higher accuracy compared with conventional CNN during training. • For datasets acquired using different scanners with different imaging protocols, transfer learning provided an efficient method to re-tune the classification model and improve accuracy.

Context Matters: Hawaiian Monk Seals Switch Between Feeding Strategies Depending on Ecological Context.

The ability to expand the behavioral repertoire past seemingly rigid morphological features enables animals to succeed in a variety of ecological contexts. The integration of morphology, performance, and behavior produces diverse animal feeding strategies. These different strategies reflect trade-offs between specialization, prey choice, and energetic expenditure, which have important consequences for understanding individual and population-level flexibility in response to environmental change. Here we examined the feeding strategies used by the Hawaiian monk seal (Neomonachus schauinslandi), an endangered marine predator. We tested how Hawaiian monk seal feeding strategies change in response to ecological context, specifically prey size and prey location at different depths. Seven captive Hawaiian monk seals were fed five prey types across a continuum of sizes, and prey were presented at three depths to represent surface, pelagic, and benthic feeding. Hawaiian monk seals used suction feeding and biting strategies, and these strategies were associated with significant differences in behavior and kinematic performance. Hawaiian monk seals used suction feeding most frequently when targeting small to medium prey (0-79% of the seal's head length) but switched to biting when consuming large prey (>80% of the seal's head length). These results demonstrate that prey size drives the transition between suction feeding and biting strategies. Seals also switched strategies based on prey position in the water column, primarily using suction feeding when prey were benthic and pelagic, and biting when prey were at the water's surface. Overall, suction feeding was three to five times faster than biting, required a smaller gape, and used fewer jaw movements, allowing seals to quickly consume numerous small to medium sized prey. In contrast, biting was slower but resulted in the ability to target larger, potentially more energy rich prey. Our results show that Hawaiian monk seals exhibit flexibility in their use of different feeding strategies, which likely facilitates increased foraging success when feeding in spatially and temporally dynamic marine environments.

Marine Protection Induces Morphological Variation in the California Moray, Gymnothorax mordax.

The effectiveness of marine protected areas (MPAs) on the general health and conservation of species, habitats, and community interactions is of great interest to researchers, managers, and recreationalists. However, the ecological and behavioral diversity of vertebrate predators of southern California kelp forests limits our ability to make general conclusions about MPA effectiveness across a variety of species. Identifying and studying species with extreme feeding habits or prey-capture strategies may offer greater insight into predator-prey relationships and reveal the trophic importance of an animal in the larger community. Moray eels (family Muraenidae) have been shown to have morphological and behavioral adaptations that allow them to consume large prey whole, identifying them as important predators. From 2015 to 2018, we studied the health and feeding behavior of a long-lived, elusive, and benthic kelp forest predator, the California moray eel (Gymnothorax mordax). We trapped eels inside and outside of Blue Cavern Onshore State Marine Conservation Area, an MPA on the northwest side of Santa Catalina Island, CA which prohibits the take of any species. Over 4 years, we captured 1736 eels. Overall, we found that morays were longer, older, heavier, had higher body condition, and were found in greater abundance within the MPA. Although fish comprised the majority of their summer diet, morays outside of the MPA were consuming a more diverse set of fish, while kelp bass comprised more than half of the diet for morays inhabiting the MPA. Additionally, we found that morays within the MPA had larger relative vertical gape distances (VGDs) and narrower heads. Our recapture data support the high site fidelity of morays, indicating that their diet and morphology are influenced by their local community. While the majority of morays are thriving in the MPA, as suggested by their robust sizes and longevity, high abundance appears to result in higher frequencies of cannibalism, the presence of an undescribed disease, and lower growth rates. Our results suggest that the MPA affects the life history of morays and may select for an alternative feeding strategy in which eels develop larger VGDs, smaller adductor muscles, and a specialized diet which is presumably influenced by the local environment. In addition, observations of cannibalistic behavior and species-specific disease provide us with important insight into natural factors that may still regulate populations removed from anthropogenic disturbances such as fishing.