Patient perspectives on the use of artificial intelligence in prostate cancer diagnosis on MRI.

OBJECTIVES: This study investigated patients' acceptance of artificial intelligence (AI) for diagnosing prostate cancer (PCa) on MRI scans and the factors influencing their trust in AI diagnoses. MATERIALS AND METHODS: A prospective, multicenter study was conducted between January and November 2023. Patients undergoing prostate MRI were surveyed about their opinions on hypothetical AI assessment of their MRI scans. The questionnaire included nine items: four on hypothetical scenarios of combinations between AI and the radiologist, two on trust in the diagnosis, and three on accountability for misdiagnosis. Relationships between the items and independent variables were assessed using multivariate analysis. RESULTS: A total of 212 PCa suspicious patients undergoing prostate MRI were included. The majority preferred AI involvement in their PCa diagnosis alongside a radiologist, with 91% agreeing with AI as the primary reader and 79% as the secondary reader. If AI has a high certainty diagnosis, 15% of the respondents would accept it as the sole decision-maker. Autonomous AI outperforming radiologists would be accepted by 52%. Higher educated persons tended to accept AI when it would outperform radiologists (p < 0.05). The respondents indicated that the hospital (76%), radiologist (70%), and program developer (55%) should be held accountable for misdiagnosis. CONCLUSIONS: Patients favor AI involvement alongside radiologists in PCa diagnosis. Trust in AI diagnosis depends on the patient's education level and the AI performance, with autonomous AI acceptance by a small majority on the condition that AI outperforms a radiologist. Respondents held the hospital, radiologist, and program developers accountable for misdiagnosis in descending order of accountability. CLINICAL RELEVANCE STATEMENT: Patients show a high level of acceptance for AI-assisted prostate cancer diagnosis on MRI, either alongside radiologists or fully autonomous, particularly if it demonstrates superior performance to radiologists alone. KEY POINTS: Prostate cancer suspicious patients may accept autonomous AI based on performance. Patients prefer AI involvement alongside a radiologist in diagnosing prostate cancer. Patients indicate accountability for AI should be shared among multiple stakeholders.

AI-assisted biparametric MRI surveillance of prostate cancer: feasibility study.

OBJECTIVES: To evaluate the feasibility of automatic longitudinal analysis of consecutive biparametric MRI (bpMRI) scans to detect clinically significant (cs) prostate cancer (PCa). METHODS: This retrospective study included a multi-center dataset of 1513 patients who underwent bpMRI (T2 + DWI) between 2014 and 2020, of whom 73 patients underwent at least two consecutive bpMRI scans and repeat biopsies. A deep learning PCa detection model was developed to produce a heatmap of all PIRADS ≥ 2 lesions across prior and current studies. The heatmaps for each patient's prior and current examination were used to extract differential volumetric and likelihood features reflecting explainable changes between examinations. A machine learning classifier was trained to predict from these features csPCa (ISUP > 1) at the current examination according to biopsy. A classifier trained on the current study only was developed for comparison. An extended classifier was developed to incorporate clinical parameters (PSA, PSA density, and age). The cross-validated diagnostic accuracies were compared using ROC analysis. The diagnostic performance of the best model was compared to the radiologist scores. RESULTS: The model including prior and current study (AUC 0.81, CI: 0.69, 0.91) resulted in a higher (p = 0.04) diagnostic accuracy than the current only model (AUC 0.73, CI: 0.61, 0.84). Adding clinical variables further improved diagnostic performance (AUC 0.86, CI: 0.77, 0.93). The diagnostic performance of the surveillance AI model was significantly better (p = 0.02) than of radiologists (AUC 0.69, CI: 0.54, 0.81). CONCLUSIONS: Our proposed AI-assisted surveillance of prostate MRI can pick up explainable, diagnostically relevant changes with promising diagnostic accuracy. KEY POINTS: • Sequential prostate MRI scans can be automatically evaluated using a hybrid deep learning and machine learning approach. • The diagnostic accuracy of our csPCa detection AI model improved by including clinical parameters.

Machine learning in the differentiation of follicular lymphoma from diffuse large B-cell lymphoma with radiomic [18F]FDG PET/CT features.

BACKGROUND: One of the challenges in the management of patients with follicular lymphoma (FL) is the identification of individuals with histological transformation, most commonly into diffuse large B-cell lymphoma (DLBCL). [18F]FDG-PET/CT is used for staging of patients with lymphoma, but visual interpretation cannot reliably discern FL from DLBCL. This study evaluated whether radiomic features extracted from clinical baseline [18F]FDG PET/CT and analyzed by machine learning algorithms may help discriminate FL from DLBCL. MATERIALS AND METHODS: Patients were selected based on confirmed histopathological diagnosis of primary FL (n=44) or DLBCL (n=76) and available [18F]FDG PET/CT with EARL reconstruction parameters within 6 months of diagnosis. Radiomic features were extracted from the volume of interest on co-registered [18F]FDG PET and CT images. Analysis of selected radiomic features was performed with machine learning classifiers based on logistic regression and tree-based ensemble classifiers (AdaBoosting, Gradient Boosting, and XG Boosting). The performance of radiomic features was compared with a SUVmax-based logistic regression model. RESULTS: From the segmented lesions, 121 FL and 227 DLBCL lesions were included for radiomic feature extraction. In total, 79 radiomic features were extracted from the SUVmap, 51 from CT, and 6 shape features. Machine learning classifier Gradient Boosting achieved the best discrimination performance using 136 radiomic features (AUC of 0.86 and accuracy of 80%). SUVmax-based logistic regression model achieved an AUC of 0.79 and an accuracy of 70%. Gradient Boosting classifier had a significantly greater AUC and accuracy compared to the SUVmax-based logistic regression (p≤0.01). CONCLUSION: Machine learning analysis of radiomic features may be of diagnostic value for discriminating FL from DLBCL tumor lesions, beyond that of the SUVmax alone.

Preoperative image-guided identification of response to neoadjuvant chemoradiotherapy in esophageal cancer (PRIDE): a multicenter observational study.

BACKGROUND: Nearly one third of patients undergoing neoadjuvant chemoradiotherapy (nCRT) for locally advanced esophageal cancer have a pathologic complete response (pCR) of the primary tumor upon histopathological evaluation of the resection specimen. The primary aim of this study is to develop a model that predicts the probability of pCR to nCRT in esophageal cancer, based on diffusion-weighted magnetic resonance imaging (DW-MRI), dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) and 18F-fluorodeoxyglucose positron emission tomography with computed tomography (18F-FDG PET-CT). Accurate response prediction could lead to a patient-tailored approach with omission of surgery in the future in case of predicted pCR or additional neoadjuvant treatment in case of non-pCR. METHODS: The PRIDE study is a prospective, single arm, observational multicenter study designed to develop a multimodal prediction model for histopathological response to nCRT for esophageal cancer. A total of 200 patients with locally advanced esophageal cancer - of which at least 130 patients with adenocarcinoma and at least 61 patients with squamous cell carcinoma - scheduled to receive nCRT followed by esophagectomy will be included. The primary modalities to be incorporated in the prediction model are quantitative parameters derived from MRI and 18F-FDG PET-CT scans, which will be acquired at fixed intervals before, during and after nCRT. Secondary modalities include blood samples for analysis of the presence of circulating tumor DNA (ctDNA) at 3 time-points (before, during and after nCRT), and an endoscopy with (random) bite-on-bite biopsies of the primary tumor site and other suspected lesions in the esophagus as well as an endoscopic ultrasonography (EUS) with fine needle aspiration of suspected lymph nodes after finishing nCRT. The main study endpoint is the performance of the model for pCR prediction. Secondary endpoints include progression-free and overall survival. DISCUSSION: If the multimodal PRIDE concept provides high predictive performance for pCR, the results of this study will play an important role in accurate identification of esophageal cancer patients with a pCR to nCRT. These patients might benefit from a patient-tailored approach with omission of surgery in the future. Vice versa, patients with non-pCR might benefit from additional neoadjuvant treatment, or ineffective therapy could be stopped. TRIAL REGISTRATION: The article reports on a health care intervention on human participants and was prospectively registered on March 22, 2018 under ClinicalTrials.gov Identifier: NCT03474341 .

Feasibility of a 2nd generation MR-compatible manipulator for transrectal prostate biopsy guidance.

OBJECTIVES: To assess the feasibility of a 2nd generation MR-compatible, remote-controlled manipulator (RCM) as an aid to perform MR-guided transrectal prostate biopsy in males with suspicion of prostate cancer (PCa). METHODS: This prospective phase I study was approved by the local ethical committee and written informed consent was obtained from each patient. Twenty patients with ≥1 cancer suspicious region (CSR) with a PI-RADS score of ≥3 detected on the diagnostic multi-parametric MRI and no prior prostate treatment underwent MR-guided biopsy with the aid of the RCM. Complications were classified according to the modified Clavien system for reporting surgical complications. For evaluation of the workflow, procedure- and manipulation times were recorded. RESULTS: All CSR's (n=20) were reachable with the MR-compatible RCM and the cancer detection rate was 70 %. The median procedure time was 36:44 minutes (range, 23 - 61 minutes) and the median manipulation time for needle guide movement was 5:48 minutes (range, 1:15 - 18:35 minutes). Two Clavien grade 1 complications were reported. CONCLUSIONS: It is feasible and safe to perform transrectal MR-guided prostate biopsy using a MR-compatible RCM as an aid. It is a fast and efficient way to biopsy suspicious prostate lesions with a minimum number of biopsies per patient. KEY POINTS: • It is feasible to perform transrectal prostate biopsy using a MR-compatible RCM. • Using a RCM for MR-guided biopsy is safe, fast, and efficient. • All cancer suspicious regions were reachable with the RCM.

Recommendations in Second Opinion Reports of Neurologic Head and Neck Imaging: Frequency, Referring Clinicians' Compliance, and Diagnostic Yield.

BACKGROUND AND PURPOSE: Second opinion reports of neurologic head and neck imaging are requested with increased regularity, and they may contain a recommendation to the clinician. Our aim was to investigate the frequency and determinants of the presence of a recommendation and the adherence by the referring physician to the recommendation in a second opinion neurology head and neck imaging report and the diagnostic yield of these recommendations. MATERIALS AND METHODS: This retrospective study included 994 consecutive second opinion reports of neurology head and neck imaging examinations performed at a tertiary care center. RESULTS: Of the 994 second opinion reports, 12.2% (121/994) contained a recommendation. An oncologic imaging indication was significantly (P = .030) associated with a lower chance of a recommendation in the second opinion report (OR = .67; 95% CI, 0.46-0.96). Clinicians followed 65.7% (88/134) of the recommendations. None of the investigated variables (patient age, sex, hospitalization status, indication for the second opinion report, experience of the radiologist who signed the second opinion report, strength of the recommendation, and whether the recommendation was made due to apparent quality issues of the original examination) were significantly associated with the compliance of the referring physician to this recommendation. The 134 individual recommendations eventually led to the establishment of 52 (38.2%) benign diagnoses and 28 (20.6%) malignant diagnoses, while no definitive diagnosis could be established in 56 (41.2%) cases. CONCLUSIONS: Recommendations are relatively common in second opinion reports of neurology head and neck imaging examinations, though less for oncologic indications. They are mostly followed by requesting physicians, thus affecting patient management. In most cases, they also lead to the establishment of a diagnosis, hence adding value to patient care.

Survival comparison of radiosurgery-eligible and -ineligible malignant glioma patients treated with hyperfractionated radiation therapy and carmustine: a report of Radiation Therapy Oncology Group 83-02.

PURPOSE: The purpose is twofold: (1) to identify the malignant glioma patients treated in a trial of hyperfractionated radiotherapy (RT) and carmustine (BCNU) who may have been eligible for a stereotactic radiosurgery (SRS) boost; and (2) to compare survival of such patients with that of those considered SRS-ineligible. PATIENTS AND METHODS: From January 1983 to July 1989, 778 malignant glioma patients were enrolled on Radiation Therapy Oncology Group (RTOG) 83-02, a randomized phase I/II hyperfractionated RT dose-escalation trial with BCNU chemotherapy. The SRS criteria used in a single-institution trial were applied to these patients; they are: Karnofsky performance status (KPS) of greater than 60; well-circumscribed tumor less than 4.0 cm; no subependymal spread; and a location not adjacent to brainstem or optic chiasm. RESULTS: Eighty-nine patients (11.9%) were identified as potentially SRS-eligible. The median survival times (MST) and 18-month survival rates of the 89 eligible and 643 ineligible patients were 14.4 versus 11.7 months and 40% versus 27%, respectively (P = .047). The MST and 18-month survival rate of the 544 SRS-ineligible patients with KPS greater than 60 were 12.1 months and 29%, respectively, and were not statistically inferior to the survival of the SRS-eligible group (P = .21). Multivariate analysis revealed age, KPS, and histopathology to be strongly predictive of survival, and SRS eligibility was also significantly predictive (P = .047). CONCLUSION: SRS-eligible patients enrolled on RTOG 83-02 had survival superior to that of the SRS-ineligible group, and this advantage is mainly due to the selection of a subgroup with a high minimum KPS.

Radiation response of human normal and leukemic hemopoietic cells assayed by in vitro colony formation.

The effect of ionizing radiation on the survival of bone marrow cells from patients with acute nonlymphocytic leukemia or from hematologically normal controls was studied using colony formation as an endpoint. A modified agar culture method which incorporated daily feeding with new medium was used to allow the growth of leukemic cell colonies. Analysis of radiation-dose survival curves revealed that normal bone marrow cell populations exhibited a relatively steep slope, with values of D0 ranging from 0.5-1.3 Gy (mean = 0.82 +/- 0.22 Gy). There was essentially no shoulder to the survival curves, with Dq values ranging from less than 0 to 0.3 Gy. The leukemic cells tested displayed survival curves that did not differ qualitatively from those obtained with normal cells, i.e., steep slopes and neglible shoulders. However, the average value of the D0 (0.62 +/- 0.15 Gy) was statistically different (p less than 0.025) than that obtained for the normal cells. The results of these studies may have implications both for the use of radiation therapy for the treatment of malignant hemopoietic diseases, and for total body irradiation prior to bone marrow transplantation.

Permanent 125iodine implants for recurrent malignant gliomas.

PURPOSE: To determine the efficacy and toxicity of permanent 125iodine implants for recurrent malignant gliomas. METHODS AND MATERIALS: Between January 1989 and January:, 59 patients with histologically confirmed recurrent malignant gliomas (22 nonglioblastoma malignant gliomas, 37 glioblastoma multiforme at the time of implant) received a permanent 125iodine implant. Patients ranged in age from 13-74 years. The median ages for the overall group, nonglioblastoma (nonGBM), and glioblastoma (GBM) groups was 47 years, 39 years, and 53 years, respectively. RESULTS: With a median follow-up of 40 months, the median survival for the 59 total patients is 1.34 years; nonGBM 2.04 years, GBM 0.9 years. Factors predictive for poor prognosis were GBM histology, age 60 years or more, target volume 17 cc or more, and/or tumor location within the corpus callosum or thalamus. Reoperations have been performed in 24 (40%) patients; 15 (25%) for tumor progression; 3 (5%) for radiation necrosis; 2 (3%) for skull necrosis/infection, and 4 (7%) for other reasons (Ommaya reservoir insertion, catheter removal, hematoma evacuation). CONCLUSION: Permanent 125iodine implants in selected patients with recurrent malignant gliomas are associated with reasonable long-term survival and a low risk of complications. Given the low incidence of radiation necrosis, future plans are to increase dose rate and/or total dose delivered with the permanent implant.

Permanent iodine-125 implants in the up-front treatment of malignant gliomas.

Between July 1989 and July 1992, 58 patients with newly diagnosed, histologically confirmed malignant gliomas (40 anaplastic astrocytomas, 18 glioblastoma multiforme) underwent implantation with low-activity iodine-125 sources. Patients were considered appropriate candidates for brachytherapy if their Karnofsky scores were > or = 70 and their contrast-enhancing tumors were < 6 cm in maximum diameter. Tumor volumes ranged from 0.1 to 90 ml. Ten patients had implants only. The other 48 patients received additional external beam radiation; 38 patients received radiation 1 to 2 weeks after the implant, and 10 patients received radiation preceding the implant. Median survival has not been reached but is currently greater than 31 months for patients with anaplastic astrocytoma and greater than 23 months for patients with glioblastoma. The rate of second operation for this group of patients was 45% (26 patients). Brain necrosis requiring resection occurred in 11 patients (19%). Although further follow-up is required, we conclude that low-activity permanent iodine-125 implants provide patients who have newly diagnosed malignant gliomas long-term survival with an acceptable risk of late complications.

Determining mucosal barrier injury to the oesophagus using CT scan.

INTRODUCTION: Oral mucositis is recognised as one of the most debilitating complications of high-dose cytostatic chemotherapy used to prepare for haematopoietic stem cell transplantation (HSCT), but very little is known about oesophageal mucositis, as endoscopy is not routinely performed. MATERIALS AND METHODS: We incorporate the computed tomography (CT) scan in the diagnostic workup of fever during neutropenia to detect evidence of pulmonary complications. This allowed us to evaluate whether mucosal barrier injury to the oesophagus can be determined. We selected 46 patients without oesophageal cancer or immune suppression (controls), who had a normal oesophagus, and measured the mucosal thickness at the upper part (UP), middle part (MP) and lower part (LP) of the oesophagus. Next, we selected 30 patients having a CT scan done for diagnostic purposes within 14 days after HSCT and measured mucosal thickness at the same levels. We also scored oral mucositis and gut toxicity. RESULTS: The mucosal thickness of the UP, MP and LP, respectively, for the controls (mean +/- SD) was 4.1 mm (+/-1.1), 4.2 mm (+/-1.2) and 4.8 mm (+/-1.3), and the corresponding values for the subjects were 5.9 mm (+/-2.2), 5.9 mm (+/-2.0) and 7.7 mm (+/-3.0). Analysis of variance showed statistically significant differences between subjects and controls at all oesophageal levels. All patients suffered from severe oral mucositis at the time. CONCLUSION: Hence, mucosal barrier injury to the oesophagus can be objectively measured using CT scan.

Permanent iodine-125 implant and external beam radiation therapy for the treatment of malignant brain tumors.

Survival data of 114 patients treated for malignant brain tumors with 125I interstitial radiation therapy at Henry Ford Hospital, Detroit, Mich. (1986-1990), are presented. The first 64 patients were treated with temporary 125I implants with a total prescribed dose of 60 Gy at a dose rate of 40 cGy/h. In order to reduce the risk of injury to the surrounding normal tissue associated with high-dose brachytherapy, a new approach was initiated using permanent implants with a lower dose rate; 50 patients were treated after surgical resection with permanent implantation of 125I seeds at a lower dose rate of 4-7 cGy/h, with a total dose of 10,000-12,000 cGy, and concurrent external radiation therapy of 5,000 cGy. The rationale of this protocol was to increase the effectiveness of the low-dose-rate implant by a concurrent 'daily' boost of external radiation, thus inhibiting the proliferation of tumor cells during the protracted low-dose radiation treatment. Survival was compared between groups with permanent and temporary implants in terms of effectiveness in tumor control as well as impact on clinical condition. Low-dose-rate implant with concurrent external radiation therapy seems to offer the best chance for long-term survival without deterioration in the clinical condition.

Multiplanar CT-guided stereotaxis and 125I interstitial radiotherapy. Image-guided tumor volume assessment, planning, dosimetric calculations, stereotactic biopsy and implantation of removable catheters.

A method using Multiplanar CT-guided stereotactic biopsy and high-dose 125I interstitial radiotherapy in patients with malignant nonresectable or recurrent brain tumors is presented. Optimal interstitial radiotherapy requires careful preoperative planning, computer-assisted dosimetry, CT-guided stereotactic biopsy and implantation of catheters that will be loaded with 125I seeds. A method is presented by which the isodose curve distribution is adjusted to the tumor size, volume and axis, allowing treatment of the imaged and histologically determined border of the tumor with 60 Gy at a dose rate of 40 rad/h.

Tumour recurrence vs radionecrosis: an indication for multitrajectory serial stereotactic biopsies.

External RT has been proved to be an important adjuvant to surgery in the treatment of malignant glioma. It has also been demonstrated, that its effect on survival is dose-dependent, although accompanied by a higher morbidity. Intents to localize the field of high dose RT to the tumour area have been performed with the aim to spare damage of the normal brain tissue. Between August 1983 to December 1987, 40 patients with malignant astrocytoma (16 GM, 24 AA) underwent high dose localized hyperfractionated external RT after surgical resection. Patients received 57.6 Gy to the tumour and oedema area associated with a boost localized to the tumour of 7.4, 14.4 or 24 Gy. In the follow-up, 16 patients died with evidence of increase in size of lesion diagnosed by CT/MRI. Since July 1987, 12 patients with recurrence or increase on size of CT/MRI lesion have undergone multitrajectory serial stereotactic biopsies. From the biopsies 8 patients were histologically diagnosed was compatible with radionecrosis. From the 4 recurrences, 2 patients were treated with 125I implants and 1 with new resection. Patients with radionecrosis were treated with corticoides and diuretics, obtaining partial or complete remission of symptoms and decrease in size of CT lesion. Undoubtly, Multiplanar/3D multitrajectory serial stereotactic biopsies play a major role in the follow-up of these patients, and accurate diagnosis need to be established for further treatment therapy. The question remains if these localized boost should be replaced by 3D Multiplanar stereotactic interstitial radiotherapy boost after surgery and conventional radiotherapy.

Lack of recovery from radiation-induced sublethal damage in human haematopoietic cells.

The effect of ionizing radiation on the growth of bone marrow cells from haematologically normal controls or from patients with acute nonlymphocytic leukaemia was studied using colony formation as an endpoint. A modified agar culture method which incorporated daily feeding with new medium was utilized to allow the growth of leukaemic cell colonies. Analysis of radiation-dose survival curves revealed that normal bone marrow cell populations exhibited a relatively steep slope, with values of D0 averaging 0.88 Gy, and essentially no shoulder to the survival curves. The leukaemic cells tested also displayed survival curves with negligible shoulders and steep slopes (average value of D0 = 0.66 Gy). The average values of D0 for normal and leukaemic cells were significantly different (P less than 0.001). The ability of hematopoietic progenitor cells to recover from the accumulation of sublethal damage was also investigated by allowing a fractionation interval between doses. As expected from the lack of significant shoulder on the survival curve, there was no evidence of recovery from sublethal damage. These results provide a rationale for the use of fractionated radiation therapy for whole-body irradiation prior to bone marrow transplant.