ABSTRACT
AIM: We investigated quantitative ultrasound (QUS) in patients with node-positive head and neck malignancies for monitoring responses to radical radiotherapy (RT). MATERIALS & METHODS: QUS spectral and texture parameters were acquired from metastatic lymph nodes 24 h, 1 and 4 weeks after starting RT. K-nearest neighbor and naive-Bayes machine-learning classifiers were used to build prediction models for each time point. Response was detected after 3 months of RT, and patients were classified into complete and partial responders. RESULTS: Single-feature naive-Bayes classification performed best with a prediction accuracy of 80, 86 and 85% at 24 h, week 1 and 4, respectively. CONCLUSION: QUS-radiomics can predict RT response at 3 months as early as 24 h with reasonable accuracy, which further improves into 1 week of treatment.
ABSTRACT
BACKGROUND: This study was conducted in order to develop a model for predicting response to neoadjuvant chemotherapy (NAC) in patients with locally advanced breast cancer (LABC) using pretreatment quantitative ultrasound (QUS) radiomics. METHODS: This was a multicenter study involving four sites across North America, and appropriate approval was obtained from the individual ethics committees. Eighty-two patients with LABC were included for final analysis. Primary tumors were scanned using a clinical ultrasound system before NAC was started. The tumors were contoured, and radiofrequency data were acquired and processed from whole tumor regions of interest. QUS spectral parameters were derived from the normalized power spectrum, and texture analysis was performed based on six QUS features using a gray level co-occurrence matrix. Patients were divided into responder or nonresponder classes based on their clinical-pathological response. Classification analysis was performed using machine learning algorithms, which were trained to optimize classification accuracy. Cross-validation was performed using a leave-one-out cross-validation method. RESULTS: Based on the clinical outcomes of NAC treatment, there were 48 responders and 34 nonresponders. A K-nearest neighbors (K-NN) approach resulted in the best classifier performance, with a sensitivity of 91%, a specificity of 83%, and an accuracy of 87%. CONCLUSION: QUS-based radiomics can predict response to NAC based on pretreatment features with acceptable accuracy.
Subject(s)
Antineoplastic Combined Chemotherapy Protocols/therapeutic use , Breast Neoplasms/diagnostic imaging , Breast Neoplasms/drug therapy , Neoadjuvant Therapy , Adult , Aged , Algorithms , Breast Neoplasms/pathology , Breast Neoplasms/surgery , Canada , Chemotherapy, Adjuvant/methods , Female , Humans , Machine Learning , Male , Middle Aged , Prospective Studies , Sensitivity and Specificity , Treatment Outcome , Ultrasonography/methods , United StatesABSTRACT
BACKGROUND: Neoadjuvant chemotherapy (NAC) is the standard of care for patients with locally advanced breast cancer (LABC). The study was conducted to investigate the utility of quantitative ultrasound (QUS) carried out during NAC to predict the final tumour response in a multi-institutional setting. METHODS: Fifty-nine patients with LABC were enrolled from three institutions in North America (Sunnybrook Health Sciences Centre (Toronto, Canada), MD Anderson Cancer Centre (Texas, USA), and Princess Margaret Cancer Centre (Toronto, Canada)). QUS data were collected before starting NAC and subsequently at weeks 1 and 4 during chemotherapy. Spectral tumour parametric maps were generated, and textural features determined using grey-level co-occurrence matrices. Patients were divided into two groups based on their pathological outcomes following surgery: responders and non-responders. Machine learning algorithms using Fisher's linear discriminant (FLD), K-nearest neighbour (K-NN), and support vector machine (SVM-RBF) were used to generate response classification models. RESULTS: Thirty-six patients were classified as responders and twenty-three as non-responders. Among all the models, SVM-RBF had the highest accuracy of 81% at both weeks 1 and week 4 with area under curve (AUC) values of 0.87 each. The inclusion of week 1 and 4 features led to an improvement of the classifier models, with the accuracy and AUC from baseline features only being 76% and 0.68, respectively. CONCLUSION: QUS data obtained during NAC reflect the ongoing treatment-related changes during chemotherapy and can lead to better classifier performances in predicting the ultimate pathologic response to treatment compared to baseline features alone.
Subject(s)
Breast Neoplasms/diagnostic imaging , Breast Neoplasms/drug therapy , Drug Monitoring , Ultrasonography , Adult , Aged , Breast Neoplasms/pathology , Chemotherapy, Adjuvant , Female , Humans , Middle Aged , Multivariate Analysis , Neoadjuvant Therapy , Neoplasm Staging , ROC Curve , Support Vector Machine , Treatment OutcomeABSTRACT
Quantitative ultrasound (QUS) techniques have been demonstrated to detect cell death in vitro and in vivo. Recently, multi-feature classification models have been incorporated into QUS texture-feature analysis methods to increase further the sensitivity and specificity of detecting treatment response in locally advanced breast cancer patients. To effectively incorporate these analytic methods into clinical applications, QUS and texture-feature estimations should be independent of data acquisition systems. The study here investigated the consistencies of QUS and texture-feature estimation techniques relative to several factors. These included the ultrasound system properties, the effects of tissue heterogeneity and the effects of these factors on the monitoring of response to neoadjuvant chemotherapy. Specifically, tumour-response-detection performance based on QUS and texture parameters using two clinical ultrasound systems was compared. Observed variations in data between the systems were small and the results exhibited good agreement in tumour response predictions obtained from both ultrasound systems. The results obtained in this study suggest that tissue heterogeneity was a dominant feature in the parameters measured with the two different ultrasound systems; whereas differences in ultrasound system beam properties only exhibited a minor impact on texture features. The McNemar statistical test performed on tumour response prediction results from the two systems did not reveal significant differences. Overall, the results in this study demonstrate the potential to achieve reliable and consistent QUS and texture-based analyses across different ultrasound imaging platforms.
Subject(s)
Breast Neoplasms/diagnostic imaging , Breast Neoplasms/drug therapy , Chemotherapy, Adjuvant , Neoadjuvant Therapy , Ultrasonography, Mammary/instrumentation , Breast Neoplasms/pathology , Cell Death , Female , Humans , Sensitivity and SpecificityABSTRACT
PURPOSE: The purpose of this study was to develop computational algorithms to best determine tumor responses early after the start of neoadjuvant chemotherapy, based on quantitative ultrasound (QUS) and textural analysis in patients with locally advanced breast cancer (LABC). METHODS: A total of 100 LABC patients treated with neoadjuvant chemotherapy were included in this study. Breast tumors were scanned with a clinical ultrasound system prior to treatment, during the first, fourth and eighth weeks of treatment, and prior to surgery. QUS parameters were calculated from ultrasound radio frequency data within tumor regions. Texture features were extracted from each QUS parametric map. Patients were classified into two groups based on identified clinical/pathological response: responders and non-responders. In order to differentiate treatment responders, three multi-feature response classification algorithms, namely a linear discriminant, a k-nearest-neighbor and a nonlinear support vector machine classifier were compared. RESULTS: All algorithms distinguished responders and non-responders with accuracies ranging between 68% and 92%. In particular, support vector machine performed the best in differentiating responders from non-responders with accuracies of 78%, 90% and 92% at weeks 1, 4 and 8 after the start of treatment, respectively. The most relevant features in separating the two response groups at early stages (weeks 1and 4) were texture features and at a later stage (week 8) were mean QUS parameters, particularly ultrasound backscatter intensity-based parameters. CONCLUSION: An early stage treatment response prediction model developed by quantitative ultrasound and texture analysis combined with modern computational methods permits offering effective alternatives to standard treatment for refractory patients.
ABSTRACT
We demonstrate the clinical utility of combining quantitative ultrasound (QUS) imaging of the breast with an artificial neural network (ANN) classifier to predict the response of breast cancer patients to neoadjuvant chemotherapy (NAC) administration prior to the start of treatment. Using a 6 MHz ultrasound system, radiofrequency (RF) ultrasound data were acquired from 100 patients with biopsy-confirmed locally advanced breast cancer prior to the start of NAC. Quantitative ultrasound mean parameter intensity and texture features were computed from the tumour core and margin, and were compared to the clinical/pathological response and 5-year recurrence-free survival (RFS) of patients. A multi-parametric QUS model in conjunction with an ANN classifier predicted patient response with 96 ± 6% accuracy, and a 0.96 ± 0.08 area under the receiver operating characteristic curve (AUC), compared to 65 ± 10 % accuracy and 0.67 ± 0.14 AUC achieved using a K-Nearest Neighbour (KNN) algorithm. A separate ANN model predicted patient RFS with 85 ± 7% accuracy, and a 0.89 ± 0.11 AUC, whereas the KNN methodology achieved a 58 ± 6 % accuracy and a 0.64 ± 0.09 AUC. The application of ANN for classifying patient response based on tumour QUS features performs well in terms of predicting response to chemotherapy. The findings here provide a framework for developing personalized a priori chemotherapy selection for patients that are candidates for NAC, potentially resulting in improved patient treatment outcomes and prognosis.
ABSTRACT
AIM: We aimed to identify quantitative ultrasound (QUS)-radiomic markers to predict radiotherapy response in metastatic lymph nodes of head and neck cancer. MATERIALS & METHODS: Node-positive head and neck cancer patients underwent pretreatment QUS imaging of their metastatic lymph nodes. Imaging features were extracted using the QUS spectral form, and second-order texture parameters. Machine-learning classifiers were used for predictive modeling, which included a logistic regression, naive Bayes, and k-nearest neighbor classifiers. RESULTS: There was a statistically significant difference in the pretreatment QUS-radiomic parameters between radiological complete responders versus partial responders (p < 0.05). The univariable model that demonstrated the greatest classification accuracy included: spectral intercept (SI)-contrast (area under the curve = 0.741). Multivariable models were also computed and showed that the SI-contrast + SI-homogeneity demonstrated an area under the curve = 0.870. The three-feature model demonstrated that the spectral slope-correlation + SI-contrast + SI-homogeneity-predicted response with accuracy of 87.5%. CONCLUSION: Multivariable QUS-radiomic features of metastatic lymph nodes can predict treatment response a priori.
ABSTRACT
OBJECTIVE: A computer-assisted technology has recently been proposed for the assessment of therapeutic responses to neoadjuvant chemotherapy in patients with locally advanced breast cancer (LABC). The system, however, extracted features from individual scans in a tumor irrespective of its relation to the other scans of the same patient, ignoring the volumetric information. This study addresses this problem by introducing a novel engineered texton-based method in order to account for volumetric information in the design of textural descriptors to represent tumor scans. METHODS: A noninvasive computer-aided-theragnosis (CAT) system was developed by employing multiparametric QUS spectral and backscatter coefficient maps. The proceeding was composed of two subdictionaries: one built on the "pretreatment" and another on "week " scans, where was 1, 4, or 8. The learned dictionary of each patient was subsequently used to compute the model (histogram of textons) for each scan of the patient. Advanced machine learning techniques including a kernel-based dissimilarity measure to estimate the distances between "pretreatment" and "mid-treatment" scans as an indication of treatment effectiveness, learning from imbalanced data, and supervised learning were subsequently employed on the texton-based features. RESULTS: The performance of the CAT system was tested using statistical tests of significance and leave-one-subject-out (LOSO) classification on 56 LABC patients. The proposed texton-based CAT system indicated significant differences in changes between the responding and nonresponding patient populations and achieved high accuracy, sensitivity, and specificity in discriminating between the two patient groups early after the start of treatment, i.e., on weeks 1 and 4 of several months of treatment. Specifically, the CAT system achieved the area under curve of 0.81, 0.83, and 0.85 on weeks 1, 4, and 8, respectively. CONCLUSION: The proposed texton-based CAT system accounted for the volumetric information in "pretreatment" and "mid-treatment" scans of each patient. It was demonstrated that this attribute of the CAT system could boost its performance compared to the cases that the features were extracted from solely individual scans.
Subject(s)
Breast Neoplasms/diagnostic imaging , Breast Neoplasms/therapy , Image Interpretation, Computer-Assisted/methods , Precision Medicine/methods , Female , Humans , Neoadjuvant Therapy , UltrasonographyABSTRACT
BACKGROUND: Pathological response of breast cancer to chemotherapy is a prognostic indicator for long-term disease free and overall survival. Responses of locally advanced breast cancer in the neoadjuvant chemotherapy (NAC) settings are often variable, and the prediction of response is imperfect. The purpose of this study was to detect primary tumor responses early after the start of neoadjuvant chemotherapy using quantitative ultrasound (QUS), textural analysis and molecular features in patients with locally advanced breast cancer. METHODS: The study included ninety six patients treated with neoadjuvant chemotherapy. Breast tumors were scanned with a clinical ultrasound system prior to chemotherapy treatment, during the first, fourth and eighth week of treatment, and prior to surgery. Quantitative ultrasound parameters and scatterer-based features were calculated from ultrasound radio frequency (RF) data within tumor regions of interest. Additionally, texture features were extracted from QUS parametric maps. Prior to therapy, all patients underwent a core needle biopsy and histological subtypes and biomarker ER, PR, and HER2 status were determined. Patients were classified into three treatment response groups based on combination of clinical and pathological analyses: complete responders (CR), partial responders (PR), and non-responders (NR). Response classifications from QUS parameters, receptors status and pathological were compared. Discriminant analysis was performed on extracted parameters using a support vector machine classifier to categorize subjects into CR, PR, and NR groups at all scan times. RESULTS: Of the 96 patients, the number of CR, PR and NR patients were 21, 52, and 23, respectively. The best prediction of treatment response was achieved with the combination mean QUS values, texture and molecular features with accuracies of 78%, 86% and 83% at weeks 1, 4, and 8, after treatment respectively. Mean QUS parameters or clinical receptors status alone predicted the three response groups with accuracies less than 60% at all scan time points. Recurrence free survival (RFS) of response groups determined based on combined features followed similar trend as determined based on clinical and pathology. CONCLUSIONS: This work demonstrates the potential of using QUS, texture and molecular features for predicting the response of primary breast tumors to chemotherapy early, and guiding the treatment planning of refractory patients.
Subject(s)
Breast Neoplasms/pathology , Neoadjuvant Therapy , Adolescent , Adult , Breast Neoplasms/drug therapy , Female , Humans , Young AdultABSTRACT
High-dose radiotherapy effects are regulated by acute tumour endothelial cell death followed by rapid tumour cell death instead of canonical DNA break damage. Pre-treatment with ultrasound-stimulated microbubbles (USMB) has enabled higher-dose radiation effects with conventional radiation doses. This study aimed to confirm acute and longitudinal relationships between vascular shutdown and tumour cell death following radiation and USMB in a wild type murine fibrosarcoma model using in vivo imaging. Methods: Tumour xenografts were treated with single radiation doses of 2 or 8 Gy alone, or in combination with low-/high-concentration USMB. Vascular changes and tumour cell death were evaluated at 3, 24 and 72 h following therapy, using high-frequency 3D power Doppler and quantitative ultrasound spectroscopy (QUS) methods, respectively. Staining using in situ end labelling (ISEL) and cluster of differentiation 31 (CD31) of tumour sections were used to assess cell death and vascular distributions, respectively, as gold standard histological methods. Results: Results indicated a decrease in the power Doppler signal of up to 50%, and an increase of more than 5 dBr in cell-death linked QUS parameters at 24 h for tumours treated with combined USMB and radiotherapy. Power Doppler and quantitative ultrasound results were significantly correlated with CD31 and ISEL staining results (p < 0.05), respectively. Moreover, a relationship was found between ultrasound power Doppler and QUS results, as well as between micro-vascular densities (CD31) and the percentage of cell death (ISEL) (R2 0.5-0.9). Conclusions: This study demonstrated, for the first time, the link between acute vascular shutdown and acute tumour cell death using in vivo longitudinal imaging, contributing to the development of theoretical models that incorporate vascular effects in radiation therapy. Overall, this study paves the way for theranostic use of ultrasound in radiation oncology as a diagnostic modality to characterize vascular and tumour response effects simultaneously, as well as a therapeutic modality to complement radiation therapy.
Subject(s)
Cell Death/radiation effects , Neoplasms/pathology , Neoplasms/radiotherapy , Animals , Mice , Mice, Inbred C57BL , Microbubbles , Ultrasonic Therapy/methods , Ultrasonic Waves , Ultrasonography/methods , Xenograft Model Antitumor Assays/methodsABSTRACT
High-frequency ultrasound (>20 MHz) spectroscopy can be used to detect noninvasively DNA damage in cell samples in vitro, and in live tissue both ex vivo and in vivo. This chapter focuses on the former two aspects. Experimental evidence suggests that morphological changes that occur in cells undergoing apoptosis result in changes in frequency-dependent ultrasound backscatter. With advances in research, ultrasound spectroscopy is advancing the boundaries of fast, label-free, noninvasive DNA damage detection technology with potential use in personalized medicine and early therapy response monitoring. Depending on the desired resolution, parametric ultrasound images can be computed and displayed within minutes to hours after ultrasound examination for cell death.
Subject(s)
Brain/diagnostic imaging , DNA Damage , Ultrasonography/methods , Animals , Apoptosis , Brain/metabolism , Brain/pathology , Humans , Laryngeal Neoplasms/diagnostic imaging , Laryngeal Neoplasms/genetics , Laryngeal Neoplasms/pathology , Leukemia, Myeloid, Acute/diagnostic imaging , Leukemia, Myeloid, Acute/genetics , Leukemia, Myeloid, Acute/pathology , Male , Rats , Rats, Inbred F344 , Tumor Cells, CulturedABSTRACT
In this chapter, we describe two new methodologies: (1) application of high-frequency ultrasound spectroscopy for in vivo detection of cancer cell death in small animal models, and (2) extension of ultrasound spectroscopy to the lower frequency range (i.e., 1-10 MHz range) for the detection of cell death in vivo in preclinical and clinical settings. Experiments using tumor xenografts in mice and cancer treatments based on chemotherapy are described. Finally, we describe how one can detect cancer response to treatment in patients noninvasively early (within 1 week of treatment initiation) using low-frequency ultrasound spectroscopic imaging and advanced machine learning techniques. Color-coded images of ultrasound spectroscopic parameters, or parametric images, permit the delineation of areas of dead cells versus viable cells using high ultrasound frequencies, and the delineation of areas of therapy response in patient tumors using clinically relevant ultrasound frequencies. Depending on the desired resolution, parametric ultrasound images can be computed and displayed within minutes to hours after ultrasound examination for cell death. A noninvasive and express method of cancer response detection using ultrasound spectroscopy provides a framework for personalized medicine with regards to the treatment planning of refractory patients resulting in substantial improvements in patient survival.
Subject(s)
Apoptosis/drug effects , DNA Damage/drug effects , Ultrasonography/methods , Animals , Antineoplastic Agents/pharmacology , Breast Neoplasms/diagnostic imaging , Breast Neoplasms/drug therapy , Breast Neoplasms/genetics , Breast Neoplasms/pathology , Drug Evaluation, Preclinical , Female , Humans , Mice , Mice, SCID , Tumor Cells, Cultured , Xenograft Model Antitumor AssaysABSTRACT
Quantitative ultrasound (QUS) can probe tissue structure and analyze tumour characteristics. Using a 6-MHz ultrasound system, radiofrequency data were acquired from 56 locally advanced breast cancer patients prior to their neoadjuvant chemotherapy (NAC) and QUS texture features were computed from regions of interest in tumour cores and their margins as potential predictive and prognostic indicators. Breast tumour molecular features were also collected and used for analysis. A multiparametric QUS model was constructed, which demonstrated a response prediction accuracy of 88% and ability to predict patient 5-year survival rates (p = 0.01). QUS features demonstrated superior performance in comparison to molecular markers and the combination of QUS and molecular markers did not improve response prediction. This study demonstrates, for the first time, that non-invasive QUS features in the core and margin of breast tumours can indicate breast cancer response to neoadjuvant chemotherapy (NAC) and predict five-year recurrence-free survival.
Subject(s)
Breast Neoplasms/diagnostic imaging , Breast Neoplasms/drug therapy , Ultrasonography , Adult , Chemotherapy, Adjuvant , Female , Humans , Kaplan-Meier Estimate , Middle Aged , Treatment OutcomeABSTRACT
BACKGROUND: Diffuse optical spectroscopy (DOS) has been demonstrated capable of monitoring response to neoadjuvant chemotherapy (NAC) in locally advanced breast cancer (LABC) patients. In this study, we evaluate texture features of pretreatment DOS functional maps for predicting LABC response to NAC. METHODS: Locally advanced breast cancer patients (n=37) underwent DOS breast imaging before starting NAC. Breast tissue parametric maps were constructed and texture analyses were performed based on grey-level co-occurrence matrices for feature extraction. Ground truth labels as responders (R) or non-responders (NR) were assigned to patients based on Miller-Payne pathological response criteria. The capability of DOS textural features computed on volumetric tumour data before the start of treatment (i.e., 'pretreatment') to predict patient responses to NAC was evaluated using a leave-one-out validation scheme at subject level. Data were analysed using a logistic regression, naive Bayes, and k-nearest neighbour classifiers. RESULTS: Data indicated that textural characteristics of pretreatment DOS parametric maps can differentiate between treatment response outcomes. The HbO2 homogeneity resulted in the highest accuracy among univariate parameters in predicting response to chemotherapy: sensitivity (%Sn) and specificity (%Sp) were 86.5% and 89.0%, respectively, and accuracy was 87.8%. The highest predictors using multivariate (binary) combination features were the Hb-contrast+HbO2-homogeneity, which resulted in a %Sn/%Sp=78.0/81.0% and an accuracy of 79.5%. CONCLUSIONS: This study demonstrated that the pretreatment DOS texture features can predict breast cancer response to NAC and potentially guide treatments.
Subject(s)
Antineoplastic Combined Chemotherapy Protocols/therapeutic use , Breast Neoplasms/diagnostic imaging , Breast Neoplasms/drug therapy , Carcinoma, Ductal, Breast/diagnostic imaging , Carcinoma, Ductal, Breast/drug therapy , Carcinoma, Lobular/drug therapy , Tomography, Optical/methods , Anthracyclines/administration & dosage , Area Under Curve , Breast Neoplasms/pathology , Bridged-Ring Compounds/administration & dosage , Carcinoma, Ductal, Breast/pathology , Carcinoma, Lobular/pathology , Chemotherapy, Adjuvant , Female , Hemoglobins/metabolism , Humans , Middle Aged , Neoadjuvant Therapy , Oxygen/metabolism , Predictive Value of Tests , ROC Curve , Spectrum Analysis , Taxoids/administration & dosage , Trastuzumab/administration & dosage , Tumor BurdenABSTRACT
GOAL: In computational biology, selecting a small subset of informative genes from microarray data continues to be a challenge due to the presence of thousands of genes. This paper aims at quantifying the dependence between gene expression data and the response variables and to identifying a subset of the most informative genes using a fast and scalable multivariate algorithm. METHODS: A novel algorithm for feature selection from gene expression data was developed. The algorithm was based on the Hilbert-Schmidt independence criterion (HSIC), and was partly motivated by singular value decomposition (SVD). RESULTS: The algorithm is computationally fast and scalable to large datasets. Moreover, it can be applied to problems with any type of response variables including, biclass, multiclass, and continuous response variables. The performance of the proposed algorithm in terms of accuracy, stability of the selected genes, speed, and scalability was evaluated using both synthetic and real-world datasets. The simulation results demonstrated that the proposed algorithm effectively and efficiently extracted stable genes with high predictive capability, in particular for datasets with multiclass response variables. CONCLUSION/SIGNIFICANCE: The proposed method does not require the whole microarray dataset to be stored in memory, and thus can easily be scaled to large datasets. This capability is an important attribute in big data analytics, where data can be large and massively distributed.
Subject(s)
Algorithms , Data Interpretation, Statistical , Gene Expression Profiling/methods , Gene Expression Regulation/physiology , Models, Statistical , Proteome/metabolism , Signal Transduction/physiology , Computer Simulation , Oligonucleotide Array Sequence Analysis/methodsABSTRACT
PURPOSE: This study demonstrated the ability of quantitative ultrasound (QUS) parameters in providing an early prediction of tumor response to neoadjuvant chemotherapy (NAC) in patients with locally advanced breast cancer (LABC). METHODS: Using a 6-MHz array transducer, ultrasound radiofrequency (RF) data were collected from 58 LABC patients prior to NAC treatment and at weeks 1, 4, and 8 of their treatment, and prior to surgery. QUS parameters including midband fit (MBF), spectral slope (SS), spectral intercept (SI), spacing among scatterers (SAS), attenuation coefficient estimate (ACE), average scatterer diameter (ASD), and average acoustic concentration (AAC) were determined from the tumor region of interest. Ultrasound data were compared with the ultimate clinical and pathological response of the patient's tumor to treatment and patient recurrence-free survival. RESULTS: Multi-parameter discriminant analysis using the κ-nearest-neighbor classifier demonstrated that the best response classification could be achieved using the combination of MBF, SS, and SAS, with an accuracy of 60 ± 10% at week 1, 77 ± 8% at week 4 and 75 ± 6% at week 8. Furthermore, when the QUS measurements at each time (week) were combined with pre-treatment (week 0) QUS values, the classification accuracies improved (70 ± 9% at week 1, 80 ± 5% at week 4, and 81 ± 6% at week 8). Finally, the multi-parameter QUS model demonstrated a significant difference in survival rates of responding and non-responding patients at weeks 1 and 4 (p=0.035, and 0.027, respectively). CONCLUSION: This study demonstrated for the first time, using new parameters tested on relatively large patient cohort and leave-one-out classifier evaluation, that a hybrid QUS biomarker including MBF, SS, and SAS could, with relatively high sensitivity and specificity, detect the response of LABC tumors to NAC as early as after 4 weeks of therapy. The findings of this study also suggested that incorporating pre-treatment QUS parameters of a tumor improved the classification results. This work demonstrated the potential of QUS and machine learning methods for the early assessment of breast tumor response to NAC and providing personalized medicine with regards to the treatment planning of refractory patients.
Subject(s)
Breast Neoplasms/diagnostic imaging , Machine Learning , Ultrasonography, Mammary/methods , Adult , Aged , Antineoplastic Agents/therapeutic use , Breast Neoplasms/drug therapy , Breast Neoplasms/pathology , Chemotherapy, Adjuvant , Female , Humans , Middle Aged , Neoadjuvant Therapy , Sensitivity and SpecificityABSTRACT
PURPOSE: Pathologists often look at whole slide images (WSIs) at low magnification to find potentially important regions and then zoom in to higher magnification to perform more sophisticated analysis of the tissue structures. Many automated methods of WSI analysis attempt to preprocess the down-sampled image in order to select salient regions which are then further analyzed by a more computationally intensive step at full magnification. Although it can greatly reduce processing times, this process may lead to small potentially important regions being overlooked at low magnification. We propose a texture analysis technique to ease the processing of H&E stained WSIs by triaging clinically important regions. METHOD: Image patches randomly selected from the whole tissue area were divided into smaller tiles and Gaussian-like texture filters were applied to them. Texture filter responses from each tile were combined together and statistical measures were derived from their histograms of responses. Bag of visual words pipeline was then employed to combine extracted features from tiles to form one histogram of words per every image patch. A support vector machine classifier was trained using the calculated histograms of words to be able to distinguish between clinically relevant and irrelevant patches. RESULT: Experimental analysis on 5151 image patches from 10 patient cases (65 tissue slides) indicated that our proposed texture technique out-performed two previously proposed colour and intensity based methods with an area under the ROC curve of 0.87. CONCLUSION: Texture features can be employed to triage clinically important areas within large WSIs.
Subject(s)
Breast Neoplasms/diagnosis , Breast Neoplasms/pathology , Image Interpretation, Computer-Assisted/methods , Female , Humans , Machine Learning , ROC CurveABSTRACT
A noninvasive computer-aided-theragnosis (CAT) system was developed for the early therapeutic cancer response assessment in patients with locally advanced breast cancer (LABC) treated with neoadjuvant chemotherapy. The proposed CAT system was based on multi-parametric quantitative ultrasound (QUS) spectroscopic methods in conjunction with advanced machine learning techniques. Specifically, a kernel-based metric named maximum mean discrepancy (MMD), a technique for learning from imbalanced data based on random undersampling, and supervised learning were investigated with response-monitoring data from LABC patients. The CAT system was tested on 56 patients using statistical significance tests and leave-one-subject-out classification techniques. Textural features using state-of-the-art local binary patterns (LBP), and gray-scale intensity features were extracted from the spectral parametric maps in the proposed CAT system. The system indicated significant differences in changes between the responding and non-responding patient populations as well as high accuracy, sensitivity, and specificity in discriminating between the two patient groups early after the start of treatment, i.e., on weeks 1 and 4 of several months of treatment. The proposed CAT system achieved an accuracy of 85%, 87%, and 90% on weeks 1, 4 and 8, respectively. The sensitivity and specificity of developed CAT system for the same times was 85%, 95%, 90% and 85%, 85%, 91%, respectively. The proposed CAT system thus establishes a noninvasive framework for monitoring cancer treatment response in tumors using clinical ultrasound imaging in conjunction with machine learning techniques. Such a framework can potentially facilitate the detection of refractory responses in patients to treatment early on during a course of therapy to enable possibly switching to more efficacious treatments.
Subject(s)
Breast Neoplasms/diagnostic imaging , Breast/diagnostic imaging , Spectrum Analysis/methods , Ultrasonography/methods , Adult , Aged , Breast Neoplasms/pathology , Female , Humans , Image Interpretation, Computer-Assisted , Middle Aged , Sensitivity and SpecificityABSTRACT
Previous studies using high-frequency ultrasound have suggested that radiofrequency (RF) spectral analysis can be used to quantify changes in cell morphology to detect cell death response to therapy non-invasively. The study here investigated this at conventional-frequencies, frequently used in clinical settings. Spectral analysis was performed using ultrasound RF data collected with a clinical ultrasound platform. Acute myeloid leukemia (AML-5) cells were exposed to cisplatinum for 0-72 hours in vitro and prepared for ultrasound data collection. Preclinical in vivo experiments were also performed on AML-5 tumour-bearing mice receiving chemotherapy. The mid-band fit (MBF) spectral parameter demonstrated an increase of 4.4 ± 1.5 dBr for in vitro samples assessed 48 hours after treatment, a statistically significant change (p < 0.05) compared to control. Further, in vitro concentration-based analysis of a mixture of apoptotic and untreated cells indicated a mean change of 10.9 ± 2.4 dBr in MBF between 0% and 40% apoptotic cell mixtures. Similar effects were reproduced in vivo with an increase of 4.6 ± 0.3 dBr in MBF compared to control, for tumours with considerable apoptotic areas within histological samples. The alterations in the size of cells and nuclei corresponded well with changes measured in the quantitative ultrasound (QUS) parameters.