Biomedical Physics & Engineering Express

Prediction of cervix cancer stage and grade from diffusion weighted imaging using EfficientNet

Abstract Purpose. This study aims to introduce an innovative noninvasive method that leverages a single image for both grading and staging prediction. The grade and the stage of cervix cancer (CC) are determined from diffusion-weighted imaging (DWI) in particular apparent diffusion coefficient (ADC) maps using deep convolutional neural networks (DCNN). Methods. datasets composed of 85 patients having annotated tumor stage (I, II, III, and IV), out of this, 66 were with grade (II and III) and the remaining patients with no reported grade were retrospectively collected. The study was IRB approved. For each patient, sagittal and axial slices containing the gross tumor volume (GTV) were extracted from ADC maps. These were computed using the mono exponential model from diffusion weighted images (b-values = 0, 100, 1000) that were acquired prior to radiotherapy treatment. Balanced training sets were created using the Synthetic Minority Oversampling Technique (SMOTE) and fed to the DCNN. EfficientNetB0 and EfficientNetB3 were transferred from the ImageNet application to binary and four-class classification tasks. Five-fold stratified cross validation was performed for the assessment of the networks. Multiple evaluation metrics were computed including the area under the receiver operating characteristic curve (AUC). Comparisons with Resnet50, Xception, and radiomic analysis were performed. Results. for grade prediction, EfficientNetB3 gave the best performance with AUC = 0.924. For stage prediction, EfficientNetB0 was the best with AUC = 0.931. The difference between both models was, however, small and not statistically significant EfficientNetB0-B3 outperformed ResNet50 (AUC = 0.71) and Xception (AUC = 0.89) in stage prediction, and demonstrated comparable results in grade classification, where AUCs of 0.89 and 0.90 were achieved by ResNet50 and Xception, respectively. DCNN outperformed radiomic analysis that gave AUC = 0.67 (grade) and AUC = 0.66 (stage). Conclusion. the prediction of CC grade and stage from ADC maps is feasible by adapting EfficientNet approaches to the medical context.

External beam patient dose verification based on the integral quality monitor (IQM ® ) output signals

Abstract Background : The Integral Quality Monitor (IQM ® ) can essentially measure the integral fluence through a segment and provide real-time information about the accuracy of radiation delivery based on comparisons of measured segment signals and pre-calculated reference values. However, the present IQM chamber cannot calculate the dose in the patient. Aim : This study aims to make use of IQM field output signals to calculate the number of monitor units (MUs) delivered through an arbitrary treatment field in order to convert Monte Carlo (MC)-generated dose distributions in a patient model into absolute dose. Methods : XiO and Monaco treatment planning systems (TPSs) were used to define treatment beam portals for cervix and esophagus conformal radiotherapy as well as prostate intensity-modulated radiotherapy for the translation of patient and beam setup information from DICOM to DOSXYZnrc. The planned beams were simulated in a patient model built from actual patient CT images and each simulated integral field/segment was weighted with its MUs before summation to get the total dose in the plan. The segment beam weights (MUs) were calculated as the ratio of the open-field IQM measured signal and the calculated signal per MU extracted from chamber sensitivity maps. These are the actual MUs delivered not just MUs set. The beam weighting method was evaluated by comparing weighted MC doses with original planned doses using profile and isodose comparisons, dose difference maps, γ analysis and dose-volume histogram (DVH) data. Results : γ pass rates of up to 98% were found, except for the esophagus plan where the γ pass rate was below 45%. DVH comparisons showed good agreement for most organs, with the largest differences observed in low-density lung. However, these discrepancies can result from differences in dose calculation algorithms or differences in MUs used for dose weighting planned by the TPS and MUs calculated using IQM field output signals. To test this, a 4-field box DOSXYZnrc MC simulation weighted with planned (XiO) MUs was compared with the same simulation weighted with IQM-based MUs. Dose differences of up to 5% were found on the isocentre slice. For XiO versus MC, up to 7% dose differences were found, indicating additional error due to limitations of XiO’s superposition algorithm. Dose differences between MC Monaco and MC EGSnrc were less than 3%. Conclusions : The most valuable comparison was MC versus MC as it eliminated algorithm discrepancies and evaluated dose differences precisely according to beam weighting. For XiO TPS, care must be taken as dose differences may also arise due to limitations in XiO’s planning software, not merely due to differences in MUs. Overall, the IQM was successfully used to compute beam dose weights to accurately reconstruct the patient dose using unweighted MC beams. Our technique can be used for pre-treatment QA provided each segment output is known and an accurate linac source model is available.

Two dimensional multifractal detrended fluctuation analysis of low coherence images for diagnosis of cervical pre-cancer

Abstract We report detection of cervical pre-cancer through their low coherence images by applying two dimensional multifractal detrended fluctuation analysis. Low coherent backscattered images of pre-cancerous cervical tissue sections were captured using a common path interferometric setup. The captured images contain both depth and lateral information of the spatial variation in refractive index (RI) occurring with progression of cervical pre-cancer. A two-dimensional multifractal detrended fluctuation analysis (2D MFDFA) was applied on these low coherent images to study the variations occurring in their fractal nature. Long-range correlations were observed in the RI fluctuations and the strength of multifractality was found to be stronger for higher grades of cervical pre-cancer. A combination of derived multifractal parameters, namely, the generalized Hurst exponent and width of singularity spectrum showed clear differences among the different grades of pre-cancers. Normal, CIN-I and CIN-II were clearly discriminated by application of support vector machine (SVM) using radial Bessel function (RBF) kernel. The specificities and sensitivities between normal and CIN-I, CIN-I and CIN-II and normal and CIN-II were found to be 94%, 88% and 93%, 96% and 98%, 100% respectively.

Dosimetric analysis of cervical cancer stage IIB patients treated with volumetric modulated arc therapy using plan uncertainty parameters module of Varian Eclipse treatment planning system

Abstract Introduction . The present study aims to investigate the dosimetric and radiobiological impact of patient setup errors (PSE) on the target and organs at risk (OAR) of the cervix carcinoma stage IIB patients treated with volumetric-modulated arc therapy (VMAT) delivery technique using plan uncertainty parameters module of Varian Eclipse treatment planning system and in-house developed DVH Analyzer program. Materials and Methods . A total of 976 VMAT plans were generated to simulate the PSE in the base plan that varies from −10 mm to 10 mm in a step size of 1 mm in x– (lateral), y– (craniocaudal), and z– (anteroposterior) directions. The different OAR and tumor (PTV) volumes were delineated in each case. Various plan quality metrics, such as conformity index (CI) and homogeneity index (HI), as well as radiobiological quantities, such as tumor control probability (TCP) and normal tissue control probability (NTCP), were calculated from the DVH bands generated from the cohort of treatment plans associated with each patient case, using an in-house developed ‘DVH Analyzer’ program. The extracted parameters were statistically analyzed and compared with the base plan’s dosimetric parameters having no PSE. Results . The maximum variation of (i) 2.4%, 21.5%, 0.8%, 2.5% in D 2cc of bladder, rectum, small bowel and sigmoid colon respectively; (ii) 19.3% and 18.9% in D max of the left and right femoral heads (iii) 16.9% in D 95% of PTV (iv) 12.1% in NTCP of sigmoid colon were observed with change of PSE in all directions. TCP was found to be considerably affected for PSEs larger than 4 mm in x + , y + , z + directions and 7 mm in x - , y - and z - directions, respectively. Conclusion . This study presents the effect of PSE on TCP and NTCP for the cervix carcinoma cases treated with VMAT technique and also recommends daily image guidance to mitigate the effects of PSE.

Enhancing pap smear image classification: integrating transfer learning and attention mechanisms for improved detection of cervical abnormalities

Abstract Cervical cancer remains a major global health challenge, accounting for significant morbidity and mortality among women. Early detection through screening, such as Pap smear tests, is crucial for effective treatment and improved patient outcomes. However, traditional manual analysis of Pap smear images is labor-intensive, subject to human error, and requires extensive expertise. To address these challenges, automated approaches using deep learning techniques have been increasingly explored, offering the potential for enhanced diagnostic accuracy and efficiency. This research focuses on improving cervical cancer detection from Pap smear images using advanced deep-learning techniques. Specifically, we aim to enhance classification performance by leveraging Transfer Learning (TL) combined with an attention mechanism, supplemented by effective preprocessing techniques. Our preprocessing pipeline includes image normalization, resizing, and the application of Histogram of Oriented Gradients (HOG), all of which contribute to better feature extraction and improved model performance. The dataset used in this study is the Mendeley Liquid-Based Cytology (LBC) dataset, which provides a comprehensive collection of cervical cytology images annotated by expert cytopathologists. Initial experiments with the ResNet model on raw data yielded an accuracy of 63.95%. However, by applying our preprocessing techniques and integrating an attention mechanism, the accuracy of the ResNet model increased dramatically to 96.74%. Further, the Xception model, known for its superior feature extraction capabilities, achieved the best performance with an accuracy of 98.95%, along with high precision (0.97), recall (0.99), and F1-Score (0.98) on preprocessed data with an attention mechanism. These results underscore the effectiveness of combining preprocessing techniques, TL, and attention mechanisms to significantly enhance the performance of automated cervical cancer detection systems. Our findings demonstrate the potential of these advanced techniques to provide reliable, accurate, and efficient diagnostic tools, which could greatly benefit clinical practice and improve patient outcomes in cervical cancer screening.

Development of handheld induction heaters for magnetic fluid hyperthermia applications and in-vitro evaluation on ovarian and prostate cancer cell lines

Abstract Objective: Magnetic fluid hyperthermia (MFH) is a still experimental technique found to have a potential application in the treatment of cancer. The method aims to reach around 41 °C–47 °C in the tumor site by exciting magnetic nanoparticles with an externally applied alternating magnetic field (AMF), where cell death is expected to occur. Applying AMFs with high spatial resolution is still a challenge. The AMFs from current and prospective MFH applicators cover relatively large areas; being not suitable for patients having metallic implants near the treatment area. Thus, there will be a clinical need for smaller magnetic field applicators. To this end, a laparoscopic induction heater (LIH) and a transrectal induction heater (TRIH) were developed. Methods: Miniature ‘pancake’ coils were wound and inserted into 3D printed enclosures. Ovarian (SKOV-3, A2780) and prostate (PC-3, LNCaP) cancer cell lines were used to evaluate the instruments’ capabilities in killing cancer cells in vitro, using Synomag®-D nanoparticles as the heat mediators. NIH3T3 normal cell lines were also used with both devices to observe if these cells tolerated the conditions applied. Results: Magnetic field intensities reached by the LIH and TRIH were 42.6 kA m−1 at 326 kHz and 26.3 kA m−1 at 303 kHz, respectively. Temperatures reached in the samples were 41 °C by the LIH and 43 °C by the TRIH. Both instruments successfully accomplished killing cancer cells, with minimal effects on normal cells. Conclusion: This work presents the first line of handheld medical induction heaters and have the potential to be a complement to existing cancer therapies. Significance: These instruments could enable the development of MFH modalities that will facilitate the clinical translation of this thermal treatment.

Forecasting patient-specific dosimetric benefit from daily online adaptive radiotherapy for cervical cancer

Abstract Objective. Adaptive Radiotherapy (ART) is an emerging technique for treating cancer patients which facilitates higher delivery accuracy and has the potential to reduce toxicity. However, ART is also resource-intensive, Requiring extra human and machine time compared to standard treatment methods. In this analysis, we sought to predict the subset of node-negative cervical cancer patients with the greatest benefit from ART, so resources might be properly allocated to the highest-yield patients. Approach. CT images, initial plan data, and on-treatment Cone-Beam CT (CBCT) images for 20 retrospective cervical cancer patients were used to simulate doses from daily non-adaptive and adaptive techniques. We evaluated the coefficient of determination (R2) between dose and volume metrics from initial treatment plans and the dosimetric benefits to the Bowel V 40 Gy , Bowel V 45 Gy , Bladder D mean , and Rectum D mean from adaptive radiotherapy using reduced 3 mm or 5 mm CTV-to-PTV margins. The LASSO technique was used to identify the most predictive metrics for Bowel V 40 Gy . The three highest performing metrics were used to build multivariate models with leave-one-out validation for Bowel V 40 Gy . Main results. Patients with higher initial bowel doses were correlated with the largest decreases in Bowel V 40 Gy from daily adaptation (linear best fit R2 = 0.77 for a 3 mm PTV margin and R2 = 0.8 for a 5 mm PTV margin). Other metrics had intermediate or no correlation. Selected covariates for the multivariate model were differences in the initial Bowel V 40 Gy and Bladder D mean using standard versus reduced margins and the initial bladder volume. Leave-one-out validation had an R2 of 0.66 between predicted and true adaptive Bowel V 40 Gy benefits for both margins. Significance. The resulting models could be used to prospectively triage cervical cancer patients on or off daily adaptation to optimally manage clinical resources. Additionally, this work presents a critical foundation for predicting benefits from daily adaptation that can be extended to other patient cohorts.

Wavelet scattering transform and entropy features in fluorescence spectral signal analysis for cervical cancer diagnosis

Abstract Cervical cancer is a prevalent malignant tumor within the female reproductive system and is regarded as a prominent cause of female mortality on a global scale. Timely and precise detection of various phases of cervical cancer holds the potential to substantially enhance both the rate of successful treatment and the duration of patient survival. Fluorescence spectroscopy is a highly sensitive method for detecting the biochemical changes that arise during cancer progression. In our study, fluorescence spectral data is collected from a diverse group of 110 subjects. The potential of the scattering transform technique for the purpose of cancer detection is explored. The processed signal undergoes an initial decomposition into scattering coefficients using the wavelet scattering transform (WST). Subsequently, the scattering coefficients are subjected to computation for fuzzy entropy, dispersion entropy, phase entropy, and spectral entropy, for effectively characterizing the fluorescence spectral signals. These combined features generated through the proposed approach are then fed to 1D convolutional neural network (CNN) classifier to classify them into normal, pre-cancerous, and cancerous categories, thereby evaluating the effectiveness of the proposed methodology. We obtained mean classification accuracy of 97% using 5-fold cross-validation. This demonstrates the potential of combining WST and entropic features for analyzing fluorescence spectroscopy signals using 1D CNN classifier that enables early cancer detection in contrast to prevailing diagnostic methods.