RESUMEN
Oligometastatic patients at [18F]F-Fluorocholine (18F-choline) PET/CT may be treated with metastasis-directed therapy (MDT). The aim of this study was to combine radiomic parameters extracted from 18F-choline PET/CT and clinical data to build machine learning (ML) models able to predict MDT efficacy. METHODS: Oligorecurrent patients (≤5 lesions) at 18F-choline PET/CT and treated with MDT were collected. A per-patient and per-lesion analysis was performed, using 2-year biochemical recurrence (BCR) after MDT as the standard of reference. Clinical parameters and radiomic features (RFts) extracted from 18F-choline PET/CT were used for training five ML Models for both CT and PET images. The performance metrics were calculated (i.e., Area Under the Curve-AUC; Classification Accuracy-CA). RESULTS: A total of 46 metastases were selected and segmented in 29 patients. BCR after MDT occurred in 20 (69%) patients after 2 years of follow-up. In total, 73 and 33 robust RFTs were selected from CT and PET datasets, respectively. PET ML Models showed better performances than CT Models for discriminating BCR after MDT, with Stochastic Gradient Descent (SGD) being the best model (AUC = 0.95; CA = 0.90). CONCLUSION: ML Models built using clinical parameters and CT and PET RFts extracted via 18F-choline PET/CT can accurately predict BCR after MDT in oligorecurrent PCa patients. If validated externally, ML Models could improve the selection of oligorecurrent PCa patients for treatment with MDT.
RESUMEN
Purpose: In stereotactic body radiation therapy (SBRT) for prostate cancer, intrafraction motion is an important source of treatment uncertainty as it could not be completely smoothed through fractionation. Herein, we compared different arrangements and beam qualities for extreme hypofractionated treatments to minimize beam delivery time and so intrafractional errors. Methods: A retrospective dataset of 11 patients was used. Three volumetric modulated arc therapy (VMAT) beam arrangements were compared for a prescription dose of 40 Gy/5 fractions: two full arcs, 6 MV flattening filter free (FFF); one full arc, 6 MV FFF; one full arc, 10 MV FFF. A plan quality index was defined to compare achievement of the planning goals. Plan complexity was evaluated with the modulation factor. Dose delivery accuracy and efficiency were measured with patient-specific quality assurance plans. Results: All treatment plans fulfilled all dose objectives. No statistical differences were found both in plan quality and complexity. Very accurate dose delivery was achieved with the three arrangements, with mean γ passing rates >96.5 % (2 %/2 mm criteria). Slightly but significantly higher γ passing rates were observed with single-arc 6 MV FFF. Contrariwise, statistically significant reductions of the delivery time were obtained with single-arc geometries: the average delivery times were 1.6 min (-46.1 %) and 1.3 min (-56.2 %) for 6 and 10 MV FFF respectively. Conclusions: The high-quality, very fast and accurate dose delivery of single-arc plans confirmed the suitability of this arrangement for prostate SBRT. In particular, the significant reduction of delivery time would improve treatment robustness against intrafraction prostate motion.
RESUMEN
Multiple myeloma (MM) is a heterogeneous neoplasm accounting for the second most prevalent hematologic disorder. The identification of noninvasive, valuable biomarkers is of utmost importance for the best patient treatment selection, especially in heterogeneous diseases like MM. Despite molecular imaging with positron emission tomography (PET) has achieved a primary role in the characterization of MM, it is not free from shortcomings. In recent years, radiomics and artificial intelligence (AI), which includes machine learning (ML) and deep learning (DL) algorithms, have played an important role in mining additional information from medical images beyond human eyes' resolving power. Our review provides a summary of the current status of radiomics and AI in different clinical contexts of MM. A systematic search of PubMed, Web of Science, and Scopus was conducted, including all the articles published in English that explored radiomics and AI analyses of PET/CT images in MM. The initial results have highlighted the potential role of such new features in order to improve the clinical stratification of MM patients, as well as to increase their clinical benefits. However, more studies are warranted before these approaches can be implemented in clinical routines.
RESUMEN
AIM: To evaluate the relevance of incidental prostate [18F]FDG uptake (IPU) and to explore the potential of radiomics and machine learning (ML) to predict prostate cancer (PCa). METHODS: We retrieved [18F]FDG PET/CT scans with evidence of IPU performed in two institutions between 2015 and 2021. Patients were divided into PCa and non-PCa, according to the biopsy. Clinical and PET/CT-derived information (comprehensive of radiomic analysis) were acquired. Five ML models were developed and their performance in discriminating PCa vs non-PCa IPU was evaluated. Radiomic analysis was investigated to predict ISUP Grade. RESULTS: Overall, 56 IPU were identified and 31 patients performed prostate biopsy. Eighteen of those were diagnosed as PCa. Only PSA and radiomic features (eight from CT and nine from PET images, respectively) showed statistically significant difference between PCa and non-PCa patients. Eight features were found to be robust between the two institutions. CT-based ML models showed good performance, especially in terms of negative predictive value (NPV 0.733-0.867). PET-derived ML models results were less accurate except the Random Forest model (NPV = 0.933). Radiomics could not accurately predict ISUP grade. CONCLUSIONS: Paired with PSA, radiomic analysis seems to be promising to discriminate PCa/non-PCa IPU. ML could be a useful tool to identify non-PCa IPU, avoiding further investigations.
Asunto(s)
Fluorodesoxiglucosa F18 , Tomografía Computarizada por Tomografía de Emisión de Positrones , Masculino , Humanos , Tomografía Computarizada por Tomografía de Emisión de Positrones/métodos , Próstata/diagnóstico por imagen , Próstata/patología , Antígeno Prostático Específico , Aprendizaje Automático , Estudios RetrospectivosRESUMEN
PURPOSE: The young working group of the Italian Association of Medical and Health Physics (AIFM) designed a survey to assess the current situation of the under 35 AIFM members. METHODS: An online survey including 65 questions was designed to gather personal information, educational issues, working and research experience, and to evaluate the AIFM activities. The survey was distributed to the under 35 members between November 2022 and February 2023, through the young AIFM mailing list and social media. RESULTS: 160 answers from 230 affiliates (70%, 31 years median age) were obtained. The results highlighted that 87% of the respondents had a fixed term/permanent employment, mainly in public hospitals (58%). Regarding Medical Physicists (MPs) training, 54% of the students left their region of origin due to the training plan (40%) and the availability of scholarships (25%) in the chosen university. Most of the respondents have no Radiation Protection Expert title, while the remaining 20%, 6%, and 3% are qualified to the first, second, and third level, respectively. Several young MPs (62.2%) were involved in research activities; however, only 28% had teaching experience, mainly within their workplace (20%, safety courses), during AIFM courses (4%), or university lectures (3%). CONCLUSIONS: This survey reported the current situation of the under 35 AIFM members, highlighting the "brain drain" phenomenon from the south to the north of Italy, mainly due to the lack of post-graduate schools, scholarships, and job opportunities. The obtained results will help the future working program of the AIFM.
Asunto(s)
Física Sanitaria , Humanos , Encuestas y Cuestionarios , Física Sanitaria/educación , Italia , UniversidadesRESUMEN
The aim of this review is to provide a comprehensive overview of the existing literature concerning the applications of positron emission tomography (PET) radiomics in lung cancer patient candidates or those undergoing immunotherapy. MATERIALS AND METHODS: A systematic review was conducted on databases and web sources. English-language original articles were considered. The title and abstract were independently reviewed to evaluate study inclusion. Duplicate, out-of-topic, and review papers, or editorials, articles, and letters to editors were excluded. For each study, the radiomics analysis was assessed based on the radiomics quality score (RQS 2.0). The review was registered on the PROSPERO database with the number CRD42023402302. RESULTS: Fifteen papers were included, thirteen were qualified as using conventional radiomics approaches, and two used deep learning radiomics. The content of each study was different; indeed, seven papers investigated the potential ability of radiomics to predict PD-L1 expression and tumor microenvironment before starting immunotherapy. Moreover, two evaluated the prediction of response, and four investigated the utility of radiomics to predict the response to immunotherapy. Finally, two papers investigated the prediction of adverse events due to immunotherapy. CONCLUSIONS: Radiomics is promising for the evaluation of TME and for the prediction of response to immunotherapy, but some limitations should be overcome.
RESUMEN
PURPOSE: A photon Monte Carlo (MC) model was commissioned for flattened (FF) and flattening filter free (FFF) 6 MV beam energy. The accuracy of this model, as a single model to be used for three beam matched LINACs, was evaluated. METHODS: Multiple models were created in RayStation v.10A for three linacs equipped with Elekta "Agility" collimator. A clinically commissioned collapsed cone (CC) algorithm (GoldCC), a MC model automatically created from the CC algorithm without further optimization (CCtoMC) and an optimized MC model (GoldMC) were compared with measurements. The validation of the model was performed by following the recommendations of IAEA TRS 430 and comprised of basic validation in a water tank, validation in a heterogeneous phantom and validation of complex IMRT/VMAT paradigms using gamma analysis of calculated and measured dose maps in a 2D-Array. RESULTS: Dose calculation with the GoldMC model resulted in a confidence level of 3% for point measurements in water tank and heterogeneous phantom for measurements performed in all three linacs. The same confidence level resulted for GoldCC model. Dose maps presented an agreement for all models on par to each other with γ criteria 2%/2mm. CONCLUSIONS: The GoldMC model showed a good agreement with measured data and is determined to be accurate for clinical use for all three linacs in this study.
Asunto(s)
Planificación de la Radioterapia Asistida por Computador , Radioterapia de Intensidad Modulada , Dosificación Radioterapéutica , Planificación de la Radioterapia Asistida por Computador/métodos , Radioterapia de Intensidad Modulada/métodos , Algoritmos , Aceleradores de Partículas , Método de Montecarlo , Fantasmas de Imagen , AguaRESUMEN
Breast cancer (BC) is a heterogeneous malignancy that still represents the second cause of cancer-related death among women worldwide. Due to the heterogeneity of BC, the correct identification of valuable biomarkers able to predict tumor biology and the best treatment approaches are still far from clear. Although molecular imaging with positron emission tomography/computed tomography (PET/CT) has improved the characterization of BC, these methods are not free from drawbacks. In recent years, radiomics and artificial intelligence (AI) have been playing an important role in the detection of several features normally unseen by the human eye in medical images. The present review provides a summary of the current status of radiomics and AI in different clinical settings of BC. A systematic search of PubMed, Web of Science and Scopus was conducted, including all articles published in English that explored radiomics and AI analyses of PET/CT images in BC. Several studies have demonstrated the potential role of such new features for the staging and prognosis as well as the assessment of biological characteristics. Radiomics and AI features appear to be promising in different clinical settings of BC, although larger prospective trials are needed to confirm and to standardize this evidence.
Asunto(s)
Neoplasias de la Mama , Femenino , Humanos , Neoplasias de la Mama/diagnóstico por imagen , Inteligencia Artificial , Tomografía Computarizada por Tomografía de Emisión de Positrones , Estudios ProspectivosRESUMEN
PURPOSE: A radiomics features classifier was implemented to evaluate segmentation quality of heart structures. A robust feature set sensitive to incorrect contouring would provide an ideal quantitative index to drive autocontouring optimization. METHODS: Twenty-five cardiac sub-structures were contoured as regions of interest in 36 CTs. Radiomic features were extracted from manually-contoured (MC) and Hierarchical-Clustering automatic-contouring (AC) structures. A robust feature-set was identified from correctly contoured CT datasets. Features variation was analyzed over a MC/AC dataset. A supervised-learning approach was used to train an Artificial-Intelligence (AI) classifier; incorrect contouring cases were generated from the gold-standard MC datasets with translations, expansions and contractions. ROC curves and confusion matrices were used to evaluate the AI-classifier performance. RESULTS: Twenty radiomics features, were found to be robust across structures, showing a good/excellent intra-class correlation coefficient (ICC) index comparing MC/AC. A significant correlation was obtained with quantitative indexes (Dice-Index, Hausdorff-distance). The trained AI-classifier detected correct contours (CC) and not correct contours (NCC) with an accuracy of 82.6% and AUC of 0.91. True positive rate (TPR) was 85.1% and 81.3% for CC and NCC. Detection of NCC at this point of the development still depended strongly on degree of contouring imperfection. CONCLUSIONS: A set of radiomics features, robust on "gold-standard" contour and sensitive to incorrect contouring was identified and implemented in an AI-workflow to quantify segmentation accuracy. This workflow permits an automatic assessment of segmentation quality and may accelerate expansion of an existing autocontouring atlas database as well as improve dosimetric analyses of large treatment plan databases.
Asunto(s)
Procesamiento de Imagen Asistido por Computador , Planificación de la Radioterapia Asistida por Computador , Corazón/diagnóstico por imagen , Radiometría , Tomografía Computarizada por Rayos XRESUMEN
The manuscript aims at providing an overview of the published algorithms/automation tool for artificial intelligence applied to imaging for Healthcare. A PubMed search was performed using the query string to identify the proposed approaches (algorithms/automation tools) for artificial intelligence (machine and deep learning) in a 5-year period. The distribution of manuscript in the various disciplines and the investigated image types according to the AI approaches are presented. The limitation and opportunity of AI application in the clinical practice or in the next future research is discussed.
Asunto(s)
Inteligencia Artificial , Aprendizaje Profundo , Algoritmos , Diagnóstico por Imagen , Aprendizaje AutomáticoRESUMEN
PURPOSE: Parametric response map (PRM) analysis of functional imaging has been shown to be an effective tool for early prediction of cancer treatment outcomes and may also be well-suited toward guiding personalized adaptive radiotherapy (RT) strategies such as sub-volume boosting. However, the PRM method was primarily designed for analysis of longitudinally acquired pairs of single-parameter image data. The purpose of this study was to demonstrate the feasibility of a generalized parametric response map analysis framework, which enables analysis of multi-parametric data while maintaining the key advantages of the original PRM method. METHODS: MRI-derived apparent diffusion coefficient (ADC) and relative cerebral blood volume (rCBV) maps acquired at 1 and 3-months post-RT for 19 patients with high-grade glioma were used to demonstrate the algorithm. Images were first co-registered and then standardized using normal tissue image intensity values. Tumor voxels were then plotted in a four-dimensional Cartesian space with coordinate values equal to a voxel's image intensity in each of the image volumes and an origin defined as the multi-parametric mean of normal tissue image intensity values. Voxel positions were orthogonally projected onto a line defined by the origin and a pre-determined response vector. The voxels are subsequently classified as positive, negative or nil, according to whether projected positions along the response vector exceeded a threshold distance from the origin. The response vector was selected by identifying the direction in which the standard deviation of tumor image intensity values was maximally different between responding and non-responding patients within a training dataset. Voxel classifications were visualized via familiar three-class response maps and then the fraction of tumor voxels associated with each of the classes was investigated for predictive utility analogous to the original PRM method. Independent PRM and MPRM analyses of the contrast-enhancing lesion (CEL) and a 1 cm shell of surrounding peri-tumoral tissue were performed. Prediction using tumor volume metrics was also investigated. Leave-one-out cross validation (LOOCV) was used in combination with permutation testing to assess preliminary predictive efficacy and estimate statistically robust P-values. The predictive endpoint was overall survival (OS) greater than or equal to the median OS of 18.2 months. RESULTS: Single-parameter PRM and multi-parametric response maps (MPRMs) were generated for each patient and used to predict OS via the LOOCV. Tumor volume metrics (P ≥ 0.071 ± 0.01) and single-parameter PRM analyses (P ≥ 0.170 ± 0.01) were not found to be predictive of OS within this study. MPRM analysis of the peri-tumoral region but not the CEL was found to be predictive of OS with a classification sensitivity, specificity and accuracy of 80%, 100%, and 89%, respectively (P = 0.001 ± 0.01). CONCLUSIONS: The feasibility of a generalized MPRM analysis framework was demonstrated with improved prediction of overall survival compared to the original single-parameter method when applied to a glioblastoma dataset. The proposed algorithm takes the spatial heterogeneity in multi-parametric response into consideration and enables visualization. MPRM analysis of peri-tumoral regions was shown to have predictive potential supporting further investigation of a larger glioblastoma dataset.