Evaluating a Machine Learning Tool for the Classification of Pathological Uptake in Whole-Body PSMA-PET-CT Scans

The importance of machine learning (ML) in the clinical environment increases constantly. Differentiation of pathological from physiological tracer-uptake in positron emission tomography/computed tomography (PET/CT) images is considered time-consuming and attention intensive, hence crucial for diagnosis and treatment planning. This study aimed at comparing and validating supervised ML algorithms to classify pathological uptake in prostate cancer (PC) patients based on prostate-specific membrane antigen (PSMA)-PET/CT. Retrospective analysis of 68Ga-PSMA-PET/CTs of 72 PC patients resulted in a total of 77 radiomics features from 2452 manually delineated hotspots for training and labeled pathological (1629) or physiological (823) as ground truth (GT). As the held-out test dataset, 331 hotspots (path.:128, phys.: 203) were delineated in 15 other patients. Three ML classifiers were trained and ranked to assess classification performance. As a result, a high overall average performance (area under the curve (AUC) of 0.98) was achieved, especially to detect pathological uptake (0.97 mean sensitivity). However, there is still room for improvement to detect physiological uptake (0.82 mean specificity), especially for glands. The ML algorithm applied to manually delineated lesions predicts hotspot labels with high accuracy on unseen data and may be an important tool to assist in clinical diagnosis.

Download Full-text

Machine learning for differentiating metastatic and completely responded sclerotic bone lesion in prostate cancer: a retrospective radiomics study

British Journal of Radiology ◽

10.1259/bjr.20190286 ◽

2019 ◽

Vol 92 (1101) ◽

pp. 20190286 ◽

Cited By ~ 6

Author(s):

Emine Acar ◽

Asım Leblebici ◽

Berat Ender Ellidokuz ◽

Yasemin Başbınar ◽

Gamze Çapa Kaya

Keyword(s):

Prostate Cancer ◽

Machine Learning ◽

Texture Analysis ◽

Area Under The Curve ◽

Ct Images ◽

Metastatic Lesions ◽

Psma Pet ◽

Pet Ct ◽

Sclerotic Lesions ◽

Psma Expression

Objective:Using CT texture analysis and machine learning methods, this study aims to distinguish the lesions imaged via 68Ga-prostate-specific membrane antigen (PSMA) positron emission tomography (PET)/CT as metastatic and completely responded in patients with known bone metastasis and who were previously treated.Methods:We retrospectively reviewed the 68Ga-PSMA PET/CT images of 75 patients after treatment, who were previously diagnosed with prostate cancer and had known bone metastasis. A texture analysis was performed on the metastatic lesions showing PSMA expression and completely responded sclerotic lesions without PSMA expression through CT images. Textural features were compared in two groups. Thus, the distinction of metastasis/completely responded lesions and the most effective parameters in this issue were determined by using various methods [decision tree, discriminant analysis, support vector machine (SVM), k-nearest neighbor (KNN), ensemble classifier] in machine learning.Results:In 28 of the 35 texture analysis findings, there was a statistically significant difference between the two groups. The Weighted KNN method had the highest accuracy and area under the curve, has been chosen as the best model. The weighted KNN algorithm was succeeded to differentiate sclerotic lesion from metastasis or completely responded lesions with 0.76 area under the curve. GLZLM_SZHGE and histogram-based kurtosis were found to be the most important parameters in differentiating metastatic and completely responded sclerotic lesions.Conclusions:Metastatic lesions and completely responded sclerosis areas in CT images, as determined by 68Ga-PSMA PET, could be distinguished with good accuracy using texture analysis and machine learning (Weighted KNN algorithm) in prostate cancer.Advances in knowledge:Our findings suggest that, with the use of newly emerging software, CT imaging can contribute to identifying the metastatic lesions in prostate cancer.

Download Full-text

Analyzing different combinations of radiomics features and clinical data for treatment response prediction based on whole-body PSMA-PET-CT scans: A machine learning based approach

10.1055/s-0040-1708365 ◽

2020 ◽

Author(s):

S Moazemi ◽

A Erle ◽

M Essler ◽

T Schultz ◽

RA Bundschuh

Keyword(s):

Machine Learning ◽

Treatment Response ◽

Clinical Data ◽

Response Prediction ◽

Whole Body ◽

Ct Scans ◽

Psma Pet ◽

Pet Ct ◽

Treatment Response Prediction

Download Full-text

Evaluating a machine learning based tool for the detection of pathological hotspots in whole-body PSMA-PET-CT scans

10.1055/s-0040-1708151 ◽

2020 ◽

Author(s):

A Erle ◽

S Moazemi ◽

M Essler ◽

T Schultz ◽

RA Bundschuh

Keyword(s):

Machine Learning ◽

Whole Body ◽

Ct Scans ◽

Psma Pet ◽

Pet Ct

Download Full-text

Determination of whole-body tumour burden on [68Ga]PSMA-11 PET/CT for response assessment of [177Lu]PSMA-617 radioligand therapy: a retrospective analysis of serum PSA level and imaging derived parameters before and after two cycles of therapy

Nuklearmedizin ◽

10.1055/a-1035-9052 ◽

2019 ◽

Vol 58 (06) ◽

pp. 443-450

Author(s):

Kerstin Michalski ◽

Michael Mix ◽

Philipp T. Meyer ◽

Juri Ruf

Keyword(s):

Standardized Uptake Value ◽

Response Assessment ◽

Tracer Uptake ◽

Whole Body ◽

Castration Resistant Prostate Cancer ◽

Tumour Burden ◽

Radioligand Therapy ◽

Psma Pet ◽

Pet Ct ◽

Before And After

Abstract Aim In patients with metastasized castration-resistant prostate cancer a reliable imaging-based therapy response assessment in addition to PSA kinetics is desirable. Recently, measurements of whole-body tumour burden by [68Ga]PSMA-11 PET/CT have been reported for response assessment in oligometastasic patients. The present study investigated the association of PSMA PET derived parameters and serum PSA level before and after [177Lu]PSMA-617 radioligand therapy (RLT). Methods This retrospective study assessed whole-body PSMA tumour volume (PSMA-TV) in 10 patients with multifocal to diffuse metastases before and after 2 cycles of RLT using volume of interest (VOI) analysis. A standardized uptake value (SUV) threshold-based approach was used to semi-automatically delineate all voxels with a SUV ≥ 2.0 g/ml using the software ROVER® (ABX Radeberg, Germany). Voxels with physiological tracer uptake (e. g. kidneys) were excluded manually. Correlations between PSMA-TV and serum PSA level before and after two cycles of RLT as well as changes thereof (ΔPSMA-TV and ΔPSA, respectively) were calculated. Results Changes of ΔPSMA-TV and ΔPSA were concordant in 7 of 10 patients. Whereas a good correlation was found between PSMA-TV and PSA before RLT (ρ = 0.81, p = 0.0049), this correlation was attenuated after RLT (ρ = 0.64, p = 0.0479). Consequently, no association was found between ΔPSMA-TV and ΔPSA (ρ = 0.39, p = 0.26). Conclusion The attenuation of the correlation of PSA and PSMA-TV after RLT suggests that in patients with advanced disease the comparison of imaging based parameters such as PSMA-TV and PSA level might be useful for an adequate monitoring of treatment response.

Download Full-text

Potential of Radiolabeled PSMA PET/CT or PET/MRI Diagnostic Procedures in Gliomas/Glioblastomas

Current Radiopharmaceuticals ◽

10.2174/1874471012666191017093721 ◽

2020 ◽

Vol 13 (2) ◽

pp. 94-98 ◽

Cited By ~ 3

Author(s):

Francesco Bertagna ◽

Domenico Albano ◽

Elisabetta Cerudelli ◽

Maria Gazzilli ◽

Raffaele Giubbini ◽

...

Keyword(s):

Case Reports ◽

Case Series ◽

Whole Body ◽

Cochrane Library ◽

Whole Body Imaging ◽

Mri Imaging ◽

Psma Pet ◽

Pet Ct ◽

Diagnostic Role

Background: Radiolabeled prostate-specific membrane antigen PSMA-based PET/CT or PET/MRI is a whole-body imaging technique currently performed for the detection of prostate cancer lesions. PSMA has been also demonstrated to be expressed by the neovasculature of many other solid tumors. Objective: The aim of this review is to evaluate the possible diagnostic role of radiolabeled PSMA PET/CT or PET/MRI in patients with gliomas and glioblastomas, by summarizing the available literature data. Methods: A comprehensive literature search of the PubMed/MEDLINE, Scopus, Embase and Cochrane library databases was conducted to find relevant published articles about the diagnostic performance of radiolabeled PSMA binding agents in PET/CT or PET/MRI imaging of patients with suspected gliomas or glioblastomas. Results: Seven case reports or case series and 3 studies enrolling more than 10 patients showed that gliomas and glioblastoma are PSMA-avid tumors. Conclusion: Radiolabeled PSMA imaging seems to be useful in analyzing glioma/glioblastoma. Further studies enrolling a wider population are needed to clarify the real clinical and diagnostic role of radiolabeled PSMA in this setting and its possible position in the diagnostic flow-chart.

Download Full-text

Effects of continuous androgen deprivation therapy on tracer uptake in primary prostate cancer in PSMA PET/CT

European Urology Open Science ◽

10.1016/s2666-1683(20)32979-7 ◽

2020 ◽

Vol 19 ◽

pp. e612

Author(s):

C. Mengxia ◽

Q. Xuefeng ◽

F. Yao ◽

Z. Qing ◽

L. Danyan ◽

...

Keyword(s):

Prostate Cancer ◽

Androgen Deprivation Therapy ◽

Tracer Uptake ◽

Androgen Deprivation ◽

Primary Prostate Cancer ◽

Psma Pet ◽

Pet Ct

Download Full-text

68Ga-PSMA-PET/CT in comparison with 18F-fluoride-PET/CT and whole-body MRI for the detection of bone metastases in patients with prostate cancer: a prospective diagnostic accuracy study

European Radiology ◽

10.1007/s00330-018-5682-x ◽

2018 ◽

Vol 29 (3) ◽

pp. 1221-1230 ◽

Cited By ~ 23

Author(s):

Eva Dyrberg ◽

Helle W. Hendel ◽

Tri Hien Viet Huynh ◽

Tobias Wirenfeldt Klausen ◽

Vibeke B. Løgager ◽

...

Keyword(s):

Prostate Cancer ◽

Diagnostic Accuracy ◽

Bone Metastases ◽

Whole Body ◽

Diagnostic Accuracy Study ◽

Whole Body Mri ◽

Psma Pet ◽

Pet Ct ◽

Accuracy Study ◽

18F Fluoride

Download Full-text

Prediction of postoperative histology in patients with prostate cancer using preoperative Ga-68-PSMA-PET/CT.

Journal of Clinical Oncology ◽

10.1200/jco.2016.34.2_suppl.144 ◽

2016 ◽

Vol 34 (2_suppl) ◽

pp. 144-144

Author(s):

Martin Boegemann ◽

Axel Semjonow ◽

Hans-Joerg Breyholz ◽

Andres Jan Schrader ◽

Laura-Maria Krabbe ◽

...

Keyword(s):

Prostate Cancer ◽

Standard Uptake Value ◽

Whole Body ◽

Diagnostic Tools ◽

Likelihood Ratios ◽

Psma Pet ◽

Pet Ct ◽

Detect Prostate Cancer ◽

Sensitivity Specificity ◽

Specific Membrane

144 Background: Recently developed 68Ga labeled prostate specific membrane antigen (PSMA) ligands were introduced as diagnostic tools to detect prostate cancer (PCa), PCa relapse and metastases with high accuracy. In this study we assessed the usability of preoperative PSMA-PET/CT information on congruency of spread of PCA compared with postoperative PCa-maps derived from radical prostatectomy (RPE) specimens. Methods: We referred 6 patients with biopsy proven high risk PCa to PSMA-PET/CT prior to RPE. Whole body PET/CT (Biograph mCT with 128 slice CT, Siemens) was performed 62±8 minutes after injection of 160±31 MBq [68Ga]-PSMA-HBED-CC (DKFZ-Ga-PSMA-11) as described by routine acquisition protocol. After RPE, prostate specimens were processed in the local pathology department. Topographical analysis of extension of PCa was reconstructed from representative slides on a schematic diagram resulting in a PCa-map of the prostate. After aligning the cutting planes of the PSMA-PET/CT to the PCa-map we defined 20 segments of the prostate and the seminal vesicles. We measured the maximum standard uptake value (SUV) of PSMA activity of the respective segments and compared the concordance of PSMA-positive and -negative areas with those of PCa and no PCa on the PCa-maps. We calculated sensitivity, specificity, positive and negative likelihood ratios (LR) taking available segments into account. Results: 106/112 segments were analyzed. 8 segments were excluded due to spillover of PSMA-activity in bladder urine. All but 3 segments with no PCa on the PCa-maps showed no uptake in PSMA-PET/CT (Specificity = 92%). The sensitivity of PSMA-PET/CT for showing PCa areas was equally 92%. The positive and negative LR for PSMA-PET/CT detecting or ruling out PCa was 11.5 and 0.09, respectively. Conclusions: This preliminary proof of concept study shows that prediction of later pathologic results in RPE-specimens could be estimated by preoperative PSMA-PET/CT. With optimized acquisition protocols it may be possible to improve our preliminary results. Perspectively PSMA-PET/CT may be helpful for identifying PCa suspicious lesions prior to prostate biopsy and support decision making prior to RPE or radiation therapy.

Download Full-text