Monitoring response to neoadjuvant therapy for breast cancer in all treatment phases using an ultrasound deep learning model
1
Citation
32
Reference
10
Related Paper
Citation Trend
Abstract:
Abstract Purpose: The present study investigated whether deep learning models (DLMs) could replace traditional ultrasound measurement models for predicting pathological responses to neoadjuvant chemotherapy (NAC) for breast cancer. Methods: Data from 57 patients (443 ultrasound images) who underwent NAC followed by surgery were analyzed. A DLM was developed for accurate breast tumor ultrasound image segmentation. The predictive abilities of the DLM, manual segmentation model (MSM), and two traditional measurement models (longest axis model [LAM] and dual-axis model [DAM]) for pathological complete response (pCR) were compared using tumor size ratios and receiver operating characteristic curves. Results: The average intersection over the union value of the DLM was 0.8087. MSM showed the best performance with an area under the curve (AUC) of 0.840; DLM performance was slightly weaker with an AUC of 0.756. The AUCs of the two traditional models were 0.778 for LAM and 0.796 for DAM. There was no significant difference in AUC values of the predictive ability of the four models. Moreover, no significant difference in AUC values of ultrasound prediction was noted between each NAC cycle (p<0.05). Conclusion: Patients in the pCR group had a significantly better response than those in the non-pCR group, and ultrasonography was predictive of pCR in the early stages of NAC. DLMs can replace traditional measurements for predicting pCR.Keywords:
Area under curve
Objective To evaluate the accuracy of six laboratory tests for the diagnosis of hyperthyroidism. Methods The levels of TT3,TT4,FT3,FT4,rT3 and s-TSH in 150 patients with hyperthyroidism and 100 normal persons were measured by RIA and IRMA. Receiver operating characteristic (ROC) curve was drawn, and the area under the curve was calculated. Results The area under ROC curve of S s-TSH ,S FT3 ,S TT3 ,S FT4 ,S rT3 and S TT4 was 0.957,0.952,0.933,0.905,0.899 and 0.874 respectively. Conclusions The accuracy of s-TSH and FT3 is the best in diagnosis of hyperthyroidism according to the ROC curve evaluation.
Area under curve
Cite
Citations (0)
The receiver operating characteristic (ROC) curve represents characteristics specific to an examination (diagnostic sensitivity and specificity) and is useful for evaluation and comparison of the diagnostic accuracy. However, the ROC curve is not widely used at present. In this symposium, we showed how to draw this curve and its practical utilization, using as examples the diagnosis of the diabetic and impaired glucose tolerance group and the diagnosis of deep-seated fungal infection and acute myocardial infarction. In the ROC curve, true positive is plotted on the vertical axis and false positive on the horizontal axis. This curve is readily drown and visually shows the diagnostic accuracy that can not be clarified by histograms. The advantages of this curve are as follows. 1. Diagnostic accuracy can be compared. 2. The significance of the reference interval in diagnosis can be evaluated. 3. The diagnostic cut-off value can be determined using this curve. 4. Combined with prevalence, the diagnostic probability can be represented quantitatively. The points that require attention are differences in the ROC curve according to selection of subjects (including controls), the time factor (disease stage) and severity (disease condition). By paying attention to these points, the ROC curve can be used as a simple and useful method in laboratory diagnosis. We hope that this curve will be widely used.
Area under curve
Curve fitting
Cite
Citations (3)
To explore the clinical significance of preoperative serum CEA, CA125, and CA19-9 levels in predicting the resectability of cholangiocarcinoma. Patients with cholangiocarcinoma diagnosed by radiologic examination and admitted to the Second Affiliated Hospital of Harbin Medical University from September 1, 2011, to November 30, 2017, were retrospectively included. The relationship between the preoperative serum CEA, CA125, and CA19-9 levels and the resectability of cholangiocarcinoma was analyzed by receiver operating characteristic (ROC) curve, as well as the best cut-off point. A total of 112 met the inclusion criteria. In 50 patients with radical surgeries, the levels of preoperative serums CEA, CA125, and CA19-9 were 5.0 ± 13.9 ng/mL, 15.3 ± 11.8 U/mL, and 257.5 ± 325.6 U/mL, respectively, which were lower than those in patients with unresectable tumor. Based on the ROC curve, the ideal CA19-9 cut-off value was determined to be 1064.1 U/mL in prediction of resectability, with a sensitivity of 53.2%, a specificity of 94.0%, and the area under the ROC curve of 0.73 (P < 0.05 P < 0.05 P < 0.05
Area under curve
Clinical Significance
Cite
Citations (70)
P50 Background and Purpose: To compare the sensitivity and specificity of quantitative cerebral blood flow (qCBF) vs. time from symptom onset to the measurement of qCBF (Time) as a predictor of cerebral infarction in patients (pts.) with acute ischemic stroke. Methods: 51 pts. with acute ischemic stroke who were assessed with XeCT, CTA and CT within 24 hours of symptom onset were studied. The MCA territory was divided into anterior and posterior divisions (two divisions/pt. for a total of 102 divisions). The average qCBF for each of these divisions was calculated and initial and follow-up CT scans were read for new infarction in both divisions. 24 divisions with evidence of prior infarction on the initial CT were excluded from the analysis. This left a total of 78 divisions available for analysis. Logistic regression was used to generate receiver operating curves (ROC) for both qCBF and Time. The area under each ROC curve is reported. Results: Twenty-one of the 78 (26.9%) divisions without initial infarction on CT had evidence of new infarction on the follow-up CT. The area under the qCBF curve was 0.81 compared with an area of 0.49 under the Time curve (p=0.00025). Excluding patients receiving thrombolytic therapy, (n=11), the area under the qCBF curve was 0.799 and the area under the Time curve was 0.590 (p=0.00004). The area under the ROC curve for qCBF was significantly greater than Time in those patients studied 180 min. (qCBF=0.76, Time=0.50; p=0.01) Conclusion: Quantitative cerebral blood flow measured by XeCT is a better predictor of new infarction on follow-up CT than Time in pts. with acute ischemic stroke. This holds true for time 180 minutes.
Stroke
Area under curve
Cite
Citations (3)
Цель. Сравнение возможностей позитронно-эмиссионной томографии, совмещенной с компьютерной томографией (ПЭТ/КТ), и магнитно-резонансной томографии (МРТ) с применением стандартного критерия (размер по короткой оси ≥1,0 см) и математической модели для выявления метастатического поражения лимфатических узлов (ЛУ) при раке шейки матки (РШМ). Материалы и методы. В исследование включены данные 108 пациентов с гистологически подтвержденным диагнозом РШМ, пролеченных хирургическим методом и консервативно, которым для оценки распространенности процесса и результатов лечения выполнены МРТ и ПЭТ/КТ. Для оценки возможностей методов в выявлении метастазов анализу подверглись 220 ЛУ. Данные ПЭТ/КТ и МРТ с применением стандартного критерия и математической модели сравнивали с результатами морфологического исследования резецированных препаратов, а также с данными динамического наблюдения (медиана наблюдения составила 8 месяцев). Для оценки диагностической эффективности методов были построены ROC-кривые (Receiver Operating Characteristic curve – ROC) с расчетом площадей под ними (Area Under Curve – AUC), определены диагностическая чувствительность (ДЧ), диагностическая специфичность (ДС), диагностическая точность (ДТ), позитивное предсказательное значение (ППЗ), негативное предсказательное значение (НПЗ). Результаты. ДЧ и ДТ МРТ математической модели составили 92,0%, 93,2% соответственно, значимо превосходили ДЧ и ДТ МРТ стандартного критерия (59,9% и 74,5% соответственно) (р<0,001) и были статистически сопоставимы с ДЧ и ДТ ПЭТ/КТ 97,8% и 96,8% соответственно) (р>0,05). AUC математической модели 0,936 значимо превосходила AUC стандартного критерия – 0,793 (р<0,05) и была сопоставима с AUC ПЭТ/КТ – 0,965 (р>0,05). Для ЛУ размером меньше 1 см по короткой оси AUC математической модели 0,876 сопоставима с AUC ПЭТ/КТ – 0,948 (р>0,05). Заключение. Математическая модель имеет высокую прогностическую ценность для выявления метастазов в ЛУ, сопоставимую с ПЭТ/КТ в том числе для ЛУ размером по короткой оси меньше 1 см. Purpose. Comparison of the capabilities of Positron Emission Tomography and Computed Tomography (PET/CT) and magnetic resonance imaging (MRI) using a standard criterion (size along the short axis D ≥1.0 cm) and a mathematical model for the detection of metastatic lesions of the lymph nodes (LN) in cervical cancer (CC). Materials and methods. The study included data from 108 patients with a histologically confirmed diagnosis of СС, treated surgically and conservatively, who underwent MRI and PET/CT to assess the prevalence of the tumor and the results of treatment. To assess the capabilities of the methods in detecting metastatic lesions, 220 LNs were analyzed. PET/ CT and MRI data using a standard criterion and a mathematical model were compared with the results of a morphological study of resected preparations, as well as with dynamic follow-up data (median follow-up was 8 months). To assess the diagnostic efficiency of the methods, ROC-curves (Receiver Operating Characteristic curve – ROC) were constructed, with the calculation of the areas under them (Area Under Curve – AUC), sensitivity (Se), specificity (Sp), accuracy, positive predictive value. negative predictive value were determined. Results. Se and accuracy MRI of the mathematical model were 92.0%, 93.2%, respectively, significantly exceeded the Se and accuracy MRI of the standard criterion (59.9% and 74.5%, respectively) and were statistically comparable with Se and accuracy PET/CT (97.8% and 96.8%, respectively). The AUC of the mathematical model was 0.936 significantly exceeded the AUC of the standard criterion 0.793 (p<0.05) and was comparable to the AUC of PET/CT 0.965 (p>0.05). For LNs smaller than 1 cm along the short axis, the AUC of the mathematical model was 0.876 and was also comparable to the PET/CT AUC 0.948 (р>0.05). Conclusion. The mathematical model has a high prognostic value for detecting LN metastases, comparable to PET/CT, including for LNs with a short axis size of less than 1 cm.
Area under curve
Cervical lymph nodes
Gold standard (test)
Cite
Citations (0)
Area under curve
Breast tumor
Cite
Citations (0)
Area under curve
Cite
Citations (0)
Objective: To evaluate the role of HbA1c in T2DM diagnosis.Methods:251 patients were underwent oral glucose tolerance test(OGTT) and HbA1c measurement.The receiver operating characteristic(ROC) curve was drawing.Results:The optimal cut-point of ROC curve of HbA1c is 7.05%,with sensitivity of 91.1%,specificity of 92.8%,area under the curve(AUC) of 0.971.However,the optimal cut-point of ROC curve of FPG was 6.94mmol/L,with sensitivity of 81%,specificity of 100%,area under the curve(AUC) of 0.944.Conclusion:Compared with FPG(cut-point: 6.94 mmol/L),HbA1c(cut-point: 7.05%) has higher sensitivity but lower specificity.
Glycated haemoglobin
Cut-point
Area under curve
Glucose tolerance test
Cite
Citations (0)
P50 Background and Purpose: To compare the sensitivity and specificity of quantitative cerebral blood flow (qCBF) vs. time from symptom onset to the measurement of qCBF (Time) as a predictor of cerebral infarction in patients (pts.) with acute ischemic stroke. Methods: 51 pts. with acute ischemic stroke who were assessed with XeCT, CTA and CT within 24 hours of symptom onset were studied. The MCA territory was divided into anterior and posterior divisions (two divisions/pt. for a total of 102 divisions). The average qCBF for each of these divisions was calculated and initial and follow-up CT scans were read for new infarction in both divisions. 24 divisions with evidence of prior infarction on the initial CT were excluded from the analysis. This left a total of 78 divisions available for analysis. Logistic regression was used to generate receiver operating curves (ROC) for both qCBF and Time. The area under each ROC curve is reported. Results: Twenty-one of the 78 (26.9%) divisions without initial infarction on CT had evidence of new infarction on the follow-up CT. The area under the qCBF curve was 0.81 compared with an area of 0.49 under the Time curve (p=0.00025). Excluding patients receiving thrombolytic therapy, (n=11), the area under the qCBF curve was 0.799 and the area under the Time curve was 0.590 (p=0.00004). The area under the ROC curve for qCBF was significantly greater than Time in those patients studied < 180 minutes (qCBF=0.92, Time=0.51; p=0.02) and > 180 min. (qCBF=0.76, Time=0.50; p=0.01) Conclusion: Quantitative cerebral blood flow measured by XeCT is a better predictor of new infarction on follow-up CT than Time in pts. with acute ischemic stroke. This holds true for time < 180 minutes and > 180 minutes.
Stroke
Area under curve
Cite
Citations (0)