The purpose of this study was to assess the potential of a deep learning model to discriminate between benign and malignant breast lesions using magnetic resonance imaging (MRI) and characterize different histological subtypes of breast lesions.
Materials and methods
We developed a deep learning model that simultaneously learns to detect lesions and characterize them. We created a lesion-characterization model based on a single two-dimensional T1-weighted fat suppressed MR image obtained after intravenous administration of a gadolinium chelate selected by radiologists. The data included 335 MR images from 335 patients, representing 17 different histological subtypes of breast lesions grouped into four categories (mammary gland, benign lesions, invasive ductal carcinoma and other malignant lesions). Algorithm performance was evaluated on an independent test set of 168 MR images using weighted sums of the area under the curve (AUC) scores.
Results
We obtained a cross-validation score of 0.817 weighted average receiver operating characteristic (ROC)-AUC on the training set computed as the mean of three-shuffle three-fold cross-validation. Our model reached a weighted mean AUC of 0.816 on the independent challenge test set.
Conclusion
This study shows good performance of a supervised-attention model with deep learning for breast MRI. This method should be validated on a larger and independent cohort. 相似文献
Background and purpose — 1 in 5 patients are dissatisfied following unicompartmental or total knee arthroplasty (UKA or TKA). This may be partly explained by failing to return to desired activity post-arthroplasty. To facilitate return to desired activity, a greater understanding of predictors of return to desired activity in UKA and TKA patients is needed. We compared rates of return to desired activity 12 months following UKA vs. TKA, and identified and compared predictors of return to desired activity 12 months following UKA vs. TKA.
Patients and methods — Patients were prospectively recruited from 2 hospitals prior to undergoing UKA or primary TKA. Patients reported preoperatively the activity/activities that were limited due to their knee that they wished to return to after arthroplasty. At 12-months postoperatively, patients reported whether they had returned to these activities (‘return to desired activity’). Preoperative predictors evaluated were age, sex, BMI, education, comorbidities, pain expectations, Oxford Knee Score (OKS), UCLA Activity Score, and EQ-5D. Generalized linear models assessed the relationship between potential predictors and return-to-desired-activity.
Results — The response rate of all patients eligible for 12-month follow-up was 74%. TKA patients (n = 575) were older (mean (SD) 70 (9) vs. 67 (10)) with a greater BMI (31 (6) vs. 30 (5)) than patients undergoing UKA (n = 420). 75% of UKA and 59% of TKA patients returned to desired activity. TKA patients had a greater risk of non-return to desired activity than patients undergoing UKA (risk ratio (95% CI) 1.5 (1.2–1.8)). Predictors of non-return to desired activity following UKA were worse OKS (0.96 (0.93–0.99)), higher BMI (1.04 (1.01–1.08)), and worse expectations (1.9 (1.2–2.8)). Predictors of non-return to desired activity following TKA were worse EQ-5D (0.53 (0.33–0.85)) and worse OKS (0.98 (0.96–1.0)).
Interpretation — UKA patients were more likely to return to desired activity than TKA patients. Predictors of return to desired activity differed following UKA and TKA. Optimizing selection of arthroplasty procedure based on patient characteristics and targeting predictors of poor outcome may facilitate return to desired activity with potential to enhance postoperative satisfaction. 相似文献
Purpose: To investigate the impact of donor and recipient factors on graft survival in penetrating keratoplasty (PK).
Material and Methods: This retrospective study included 365 eyes that underwent PK using corneas from 231 donors between June 2010 and June 2015. Patients were divided into three groups (group 1: primary endothelial diseases; group 2: iatrogenic endothelial disorders; and group 3: other pathologies with a healthy endothelium) according to PK indications. The primary outcome measure was corneal graft survival at the last visit (clear or opaque). Graft clarity was assessed using Kaplan–Meier survival analysis.
Results: The most frequent PK indication was keratoconus (KC) (20.5%) followed by pseudophakic bullous keratopathy (PBK) (18.9%). Donor age had a negative impact on endothelial cell density (ECD) measured by an eye bank specular microscope (p < 0.001). Median best-corrected visual acuity in logarithm of the minimum angle of resolution units increased from 2.1 to 0.8 at 1 year after PK (p < 0.001). The clear graft rate was 96.7% at year 1, 88.8% at year 2, and 85.5% at year 3. Overall graft survival was 84.9% during a median of 39 months (range: 24–79 months) of follow-up. A higher graft survival rate (67.2%) was observed in KC compared to PBK during 6 years (p < 0.001). Recipients younger than 50 years of age showed a better graft survival rate than those older than 70 years of age (p = 0.037). Donor ECD, time between excision and death, and preservation time had no significant effect on graft survival. Frequent graft rejection episodes (GREs) and additional procedures during surgery had a negative impact on graft survival (p < 0.001 and p = 0.014, respectively). A worse graft survival was observed in group 2 compared to groups 1 and 3 (p = 0.042).
Conclusions: Young recipient age and KC were associated with a better graft survival. Graft endothelial density and preservation time had no impact on graft survival. PBK, low vision at baseline and year 1, frequent GREs, and additional interventions during surgery had a negative impact on graft survivals. 相似文献
AbstractObjectives: This study examined word use as an indicator of interpersonal positive reframing in daily conversations of couples coping with breast cancer and as a predictor of stress.Design: The Electronically Activated Recorder (EAR) and Linguistic Inquiry and Word Count (LIWC) were used to examine naturally occurring word use conceptually linked to positive reframing (positive emotion, negative emotion, and cognitive processing words).Sample: Fifty-two couples coping with breast cancer.Methods: Couples wore the EAR, a device participants wear, that audio-recorded over one weekend (>16,000 sound files), and completed self-reports of positive reframing (COPE) and stress (Perceived Stress Scale). LIWC, a software program, measured word use.Findings: Both partners’ word use (i.e., positive emotion and cognitive processing words) was associated with their own reported positive reframing, and spouses’ word use was also indicative of patients’ positive reframing. Results also revealed that, in general, words indicating positive reframing predicted lower levels of stress.Conclusions: Findings supported the hypothesis that partners—and particularly spouses of breast cancer patients—may assist each other’s coping by positively reframing the cancer experience and other negative experiences in conversation. 相似文献
To obtain an overview of the management and outcomes of children aged 18 years or younger diagnosed with differentiated thyroid carcinoma of follicular cell origin across the UK, by collecting and analysing data from the limited number of centres treating these patients. This multicentre data might provide a more realistic perspective than single-institution series.
Materials and methods
Six centres submitted data extracted from historical records on patients aged 18 years or younger, diagnosed between 1964 and 2017. The univariate and multivariable Cox proportional hazard model was used to identify potential predictors of progression-free survival, using national data as a control.
Results
Data on 166 patients were available for analysis. Females (74%) were predominant, and the age ranged from 3 to 19 years at diagnosis, mean 14.1 years. Nodal metastases were present in 51%; 12% had distant metastases. After surgery, 95% received radioactive iodine (39% on more than one occasion) and 4% received external beam radiotherapy. With a median follow-up duration of 5 years, 69% are alive with no evidence of disease; 20% are alive with a raised thyroglobulin level as the only evidence of residual disease; 6% have residual structural disease detectable on imaging; 2% have died, from cerebral metastases.
Conclusion
Despite most patients having advanced disease at presentation, outcomes are very good. A national prospective registry should allow systematic collection of good-quality data and may facilitate research to further improve outcomes. 相似文献
Malignant pleural mesothelioma (MPM) is an uncommon cancer with a poor prognosis and heterogeneous survival. Surgery for MPM is offered in some specialist centers to highly selected patients. A previously described classification and regression tree (CART) model stratified survival in unselected MPM patients using routinely collected clinical data. This study aimed to examine the performance of this CART model on a highly selected surgical population.
Methods
Data were collected from subjects undergoing cytoreductive surgery for MPM from specialist centers in Hyõgo, Japan, and Sydney, Australia, between 1991 and 2016. The CART model was applied using the combination of clinical variables to stratify subjects into risk groups (1 through 4); survival characteristics were then compared.
Results
Two hundred eighty-nine cases were included (205 from Australia, 84 from Japan). Overall median survival was 34.6 (interquartile range: 17.5–56.1) months; median age was 63.0 (interquartile range: 57.0–67.8) years, and 83.0% (n = 240) were male. There were no clinically meaningful differences between the two cohorts. Survival across the four risk groups was significantly different (p < 0.0001); the model stratified survival well with a Harrell's concordance statistic of 0.62 (95% confidence interval: 0.57–0.66) at 36 months. The group with the longest survival (median, 82.5 months) had: no weight loss, hemoglobin > 153 g/L and serum albumin > 43 g/L at time of referral to the surgical center.
Conclusions
Using routinely available clinical variables, the CART model was able to stratify surgical patients into risk groups with statistically different survival characteristics with fair to good performance. Presence of weight loss, anemia, and low albumin should confer caution when considering surgical therapy for MPM. 相似文献
In a prospective study, we sought to determine acceptability of linkage of administrative and clinical trial data among Canadian patients and Research Ethics Boards (REBs). The goal is to develop a more harmonized approach to data, with potential to improve clinical trial conduct through enhanced data quality collected at reduced cost and inconvenience for patients. On completion of the original LY.12 randomized clinical trial in lymphoma ({"type":"clinical-trial","attrs":{"text":"NCT00078949","term_id":"NCT00078949"}}NCT00078949), participants were invited to enrol in the Long-term Innovative Follow-up Extension (LIFE) component. Those consenting to do so provided comprehensive identifying information to facilitate linkage with their administrative data. We prospectively designed a global assessment of this innovative approach to clinical trial follow-up including rates of REB approval and patient consent. The pre-specified benchmark for patient acceptability was 80%. Of 16 REBs who reviewed the research protocol, 14 (89%) provided approval; two in Quebec declined due to small patient numbers. Of 140 patients invited to participate, 115 (82%, 95% CI 76 to 88%) from across 9 Canadian provinces provided consent and their full name, date of birth, health insurance number and postal code to facilitate linkage with their administrative data for long-term follow-up. Linkage of clinical trial and administrative data is feasible and acceptable. Further collaborative work including many stakeholders is required to develop an optimized secure approach to research. A more coordinated national approach to health data could facilitate more rapid testing and identification of new effective treatments across multiple jurisdictions and diseases from diabetes to COVID-19. 相似文献
Across the spectrum of patient care for opioid overdose, an important, yet frequently overlooked feature is the bystander, or witness to the overdose event. For other acute medical events such as cardiac arrest and stroke, research supports that the presence of a bystander is associated with better outcomes. Despite the similarities, however, this well-established conceptual framework has yet to be applied in the context of overdose patient outcomes. The objective of this study was to assess the association between the nature of the bystander-patient relationship and prehospital care measures in patients being treated for opioid overdose. 相似文献
locStra is an ‐package for the analysis of regional and global population stratification in whole‐genome sequencing (WGS) studies, where regional stratification refers to the substructure defined by the loci in a particular region on the genome. Population substructure can be assessed based on the genetic covariance matrix, the genomic relationship matrix, and the unweighted/weighted genetic Jaccard similarity matrix. Using a sliding window approach, the regional similarity matrices are compared with the global ones, based on user‐defined window sizes and metrics, for example, the correlation between regional and global eigenvectors. An algorithm for the specification of the window size is provided. As the implementation fully exploits sparse matrix algebra and is written in C++, the analysis is highly efficient. Even on single cores, for realistic study sizes (several thousand subjects, several million rare variants per subject), the runtime for the genome‐wide computation of all regional similarity matrices does typically not exceed one hour, enabling an unprecedented investigation of regional stratification across the entire genome. The package is applied to three WGS studies, illustrating the varying patterns of regional substructure across the genome and its beneficial effects on association testing. 相似文献
