What’s next for VI-RADS? Updates and future perspectives from the ACR VI-RADS steering committee Authors: Sungmin Woo, Valdair F Muglia, Lyndon Luk, Mitsuru Takeuchi, Maarten de Rooij, Bernadette Redd, Samuel J Galgano, Jason Efstathiou, Alberto Briganti, J Alfred Witjes, Valeria Panebianco, Hebert Alberto Vargas Published: 2026 ABSTRACT AbstractSince the introduction of the Vesical Imaging-Reporting and Data System (VI-RADS), MRI has become an important imaging modality in the management of patients with bladder cancer. Its excellent diagnostic performance for determining muscle invasion in bladder cancer has been supported by numerous prospective and retrospective studies. Nevertheless, there needs to be continued improvement of the diagnostic performance of VI-RADS and sustained efforts to address remaining unmet clinical needs within the field of bladder cancer. In this paper, we highlight several such areas, some of which are actively being investigated. These include (1) whether to use intravenous contrast media or not (multiparametric vs. biparametric MRI), (2) quantitative metrics to enhance assessment of muscle invasion, (3) anatomic locations that come with pitfalls in staging, (4) introduction of bladder MRI image quality, (5) neoadjuvant chemotherapy VI-RADS (nacVI-RADS) and other considerations needed in the treatment response assessment after systemic therapies, (6) need for wider adoption, training, and implementation, and (7) application of artificial intelligence. Introduction Bladder cancer is globally the 9th most commonly diagnosed cancer, with an estimated 614,298 new cases and 220,596 deaths in 2022 [1]. The distinction between non-muscle-invasive bladder cancer (NMIBC) and muscle-invasive bladder cancer (MIBC) is critical in that they are substantially different with regards to management and prognosis. NMIBC is generally treated with transurethral resection of bladder tumor (TURBT) with or without subsequent intravesical therapy. On the other hand, MIBC is managed with radical cystectomy preferentially after a neoadjuvant therapy or alternatively trimodal therapy–a combination of TURBT, radiation, and chemotherapy with bladder sparing [2–4]. Traditionally, TURBT and computed tomography had been used to determine the locoregional extent of bladder cancer in order to guide subsequent management. However, since the introduction of the Vesical Imaging-Reporting and Data System (VI-RADS), a substantial body of prospective and retrospective evidence has established its excellent diagnostic performance in determining muscle invasion in bladder cancer [5]. The assessment of muscle invasiveness of bladder cancer using VI-RADS is based on the following 3 key MRI sequences: T2-weighted imaging, diffusion-weighted imaging (DWI), and dynamic contrast-enhanced MRI (DCE-MRI). The criteria for each of MRI sequence and the algorithmic approach for establishing the final VI-RADS score have been extensively provided in the literature [5–8]. This standardized framework has streamlined multiparametric MRI (mpMRI) acquisition, the assessment of muscle invasion probability, and the reporting of findings using a five-point scale, which is summarized in Table 1. As a result, the latest European Association of Urology (EAU) guidelines now recommend MRI to be included as a pretreatment assessment modality for assessing local disease extent in bladder cancer [3, 4]. Table 1Summary of VI-RADS categoriesOverall VI-RADS categoryDefinitionScoring of individual MRI sequences1Muscle invasion highly unlikelySC = 1, CE = 1, and DW = 12Muscle invasion unlikelySC = 2 or 3, CE = 2, and DW = 23Muscle invasion equivocalSC = 3, CE & DW = 2 or 3 (but not both 2)4Muscle invasion likelySC = 4, CE = 4, and DW = 4SC = 4, CE & DW = 3 or 4 (but not both 3)SC = 3 or 5, DW = 4, and CE = 45Muscle invasion (or invasion beyond bladder wall) very likelySC = 5, CE & DW = 4 or 5 (but not both 4)CE = contrast-enhanced category; DW = diffusion-weighted category; SC = structural category based on T2-weighted imaging Summary of VI-RADS categories SC = 4, CE = 4, and DW = 4 SC = 4, CE & DW = 3 or 4 (but not both 3) SC = 3 or 5, DW = 4, and CE = 4 CE = contrast-enhanced category; DW = diffusion-weighted category; SC = structural category based on T2-weighted imaging VI-RADS: Diagnostic performance and the latest updates Several systematic reviews and meta-analyses have summarized the diagnostic performance of VI-RADS for evaluating muscle invasion in bladder cancer [9–11]. The most recent one was published in 2024, evaluating a total of 31 studies with 3,798 bladder cancer patients, and demonstrated that the pooled sensitivity and specificity using a cutoff of VI-RADS score of ≥ 3 were 89% (95% confidence interval [CI]: 86–91) and 89% (95% CI: 85–91), respectively [11]. A slightly older meta-analysis published in 2022, in which 20 studies and 2,477 patients were included, additionally demonstrated that a higher cutoff (VI-RADS score of ≥ 4 vs. ≥ 3) can be used to obtain higher specificity [10]: for VI-RADS ≥ 4, sensitivity and specificity of 78% (95% CI: 74–81) and 94% (95% CI: 91–96) vs. for VI-RADS ≥ 3, sensitivity and specificity of 87% (95% CI: 82–91) and 86% (95% CI: 80–90). However, the equivocal nature of VI-RADS 3 remains a concern and the proportion of MIBC within this category ranges from 34% to 83% across the literature [12–15]. The inter-reader agreement of VI-RADS has been demonstrated to be excellent. Specifically, in a systematic review and meta-analysis of 19 studies consisting of 2,439 participants, overall agreement across all VI-RADS was Cohen’s kappa of 0.76 (95% CI: 0.73–0.80) [16]. Further studies have shown that the application of VI-RADS can be expanded to assess likelihood of extravesical extension [17] and decide on the need for a repeat TURBT [18]. Most recently, the Prospective BladderPath Trial confirmed the feasibility of the “mpMRI-directed pathway”, in which patients with suspected bladder cancer undergo a flexible cystoscopy and mpMRI (with MRI sequences and reporting that were compatible with VI-RADS) instead of only the traditional TURBT [19]. 92% of the patients with suspected MIBC underwent mpMRI which led to a 45-day decrease in the time to receiving appropriate treatment. Future perspectives and research directions While it is clear that the introduction of VI-RADS has accelerated the standardization and wider application of bladder MRI for assessing the locoregional extent of bladder cancer, there needs to be continued assessment of its diagnostic performance and sustained effort to improve and address unmet clinical needs [6, 20]. In this section, we highlight several such areas, some of which are actively being investigated. “To contrast or not to contrast–that is the question”: mpMRI vs. bpMRI The current standard for assessment of muscle invasion in bladder cancer based on VI-RADS is to use intravenous contrast media (i.e., mpMRI). However, there is growing interest in omitting it (i.e., biparametric MRI [bpMRI]) as this approach can provide several advantages. Not only does bpMRI shorten acquisition time, it does not require intravenous access, improving patient comfort. Perhaps the biggest question is whether it retains similar diagnostic performance compared to mpMRI. The latest prospective study showed that they were not significantly different: sensitivity and specificity were 76% (95% CI: 53–92) and 93% (95% CI: 86–97) for bpMRI vs. 86% (95% CI: 64–97) and 95% (95% CI: 89–98) for mpMRI [21]. Nevertheless, it is unclear whether this similar diagnostic performance is applicable to readers across different levels of experience. In addition, it has been suggested that bpMRI may have lower sensitivity than mpMRI when evaluating muscle invasion in bladder cancers with variant histology [22]. Furthermore, DCE-MRI remains an important safety net as DWI is prone to artifacts (e.g., motion from bladder and bowel peristalsis and susceptibility from rectal gas). For example, features on DCE-MRI such as peritumoral enhancement which have been suggested to be specific for muscle invasion, if present, could add to the diagnostic confidence for predicting MIBC when artifacts render DWI suboptimal [23]. Larger-scale studies may potentially guide whether bpMRI can be a suitable alternative to the current mpMRI standard or establish in which subset of patients contrast media can be safely omitted, such as how recent efforts such as the PRIME trial have shown this for prostate MRI [24]. Figure 1 shows a representative example in which using intravenous contrast (mpMRI) provided incremental imaging features to more confidently predict NMIBC compared with bpMRI. Dynamic contrast-enhanced (DCE) MRI improving prediction of non-muscle-invasive bladder cancer in 78-year-old man. (A) T2-weighted imaging, (B) diffusion-weighted imaging, and (C) apparent diffusion coefficient map show a broad-based polypoid mass (arrow) in the left posterolateral bladder wall measuring 1.4 × 1.4 cm. No stalk or thickened inner layer is present and there is a smooth interface between tumor and the muscle layer. (D) Dynamic contrast-enhanced MRI shows a delayed enhancing stalk (arrowhead) consistent with a DCE category 2. VI-RADS score would be 3 based on biparametric MRI and 2 based on multiparametric MRI. Transurethral resection revealed non-muscle-invasive bladder cancer “Images are more than pictures, they are data”: Quantitative metrics to enhance assessment The assessment of bladder MRI using VI-RADS is primarily based on subjective and qualitative interpretation (other than the size criteria of 1 cm for differentiating VI-RADS 1 and 2 lesions). This certainly is pragmatic and also has been proven to show good diagnostic performance. However, there is increasing interest in incorporating quantitative metrics with the hopes of further improving the diagnostic performance and inter-reader agreement of VI-RADS. Some of the key examples of quantitative imaging features include apparent diffusion coefficient (ADC) values derived from DWI and tumor-contact length (TCL) [13, 25–28]. For example, one study showed that adding ADC to VI-RADS increased the area under the curve from 0.886 to 0.942 [28]. In another study that developed an imaging-based nomogram to predict MIBC, TCL was shown to substantially contribute to the probability when compared with the VI-RADS score [14]. TCL cutoffs of > 2 cm, > 2.5 cm, and > 3 cm have been proposed by various investigators for predicting MIBC [14, 26, 29]. A representative example where application of TCL was helpful in downgrading suspicion for muscle invasion is shown in Fig. 2. While these quantitative imaging features have potential to improve the diagnostic performance of the current version of VI-RADS, it must be emphasized that incorporation of such would require universal adherence to the technical requirements stated in the VI-RADS framework. Even with such, there is substantial variability in ADC values that can stem from multiple factors such as technical acquisition parameters (e.g., b-values) or scanner vendor and model which needs to be taken into consideration when interpreting and applying ADC values in the context of MRI for bladder cancer assessment [30]. Tumor contact length used in conjunction with VI-RADS to correctly identify non-muscle-invasive bladder tumor in 77-year-old man. (A) T2-weighted imaging, (B) diffusion-weighted imaging, (C) apparent diffusion coefficient map, and (D) dynamic contrast-enhanced MRI show 1.7 × 0.9 cm polypoid tumor with broad base (arrow) at the left posterolateral wall. No stalk or thickened inner layer is seen resulting in a VI-RADS score of 3. (A) Tumor contact length measured on (A) T2-weighted imaging is 1.6 cm, less than several proposed cutoffs (e.g., 2.0 cm, 2.5 cm, 3.0 cm) for differentiating muscle-invasive and non-muscle-invasive bladder tumors. Transurethral resection revealed non-muscle-invasive bladder cancer “Location, location, location”: Addressing anatomic pitfalls in staging The strength of MRI for evaluating muscle invasion lies in its ability to differentiate the tumor from various anatomical structures (i.e., muscle layer, perivesical fat), each with their combination of unique signal intensity characteristics on mpMRI. However, certain anatomical locations make this challenging due to inherent morphological complexity of the muscle layer, including the ureterovesical junction, bladder neck, and bladder diverticula. For example, the ureterovesical junction by definition is where the ureter inserts into the bladder wall, and therefore a physiological defect (i.e., lumen of the ureter) is present. The bladder neck also has a similar issue with the bladder lumen continuing on to the prostatic urethra, which can be further complicated by intravesical protrusion of benign prostatic hyperplasia of the prostate in males. Bladder diverticula do not have the muscularis propria layer and therefore there can be no “muscle-invasive disease” but rather jumps straight to perivesical invasion (Fig. 3). There have been a few studies directly addressing these special scenarios [15, 31]. Nevertheless, more work is needed to establish whether there should be modifications or supplementary rules for using VI-RADS in these anatomical regions. 74-year-old man with intradiverticular bladder tumor illustrating difficulty of applying VI-RADS. (A) T2-weighted imaging demonstrates 3.1 × 2.5 cm tumor (*) with intermediate T2 signal occupying right posterolateral bladder diverticulum (arrowheads). (B) Tumor (*) shows marked degree of restricted diffusion on diffusion-weighted imaging. There is loss of normal T2 dark signal bladder wall with irregular margin with the perivesical fat (arrow). As there is no “muscularis propria” layer in bladder diverticula, this represents a direct jump from non-muscle-invasive bladder tumor to extravesical extension. Despite resection and chemotherapy, patient died from progression of disease (not shown) after 2 years “Garbage in, garbage out”: standardizing and improving bladder MRI quality While VI-RADS has established standards for acquisition of MRI for staging bladder cancer, the quality of MRI examinations which are used to apply VI-RADS is variable in clinical practice. This may be attributed to several factors including the fact that VI-RADS is not widely adopted and there had been no formal method for assessing the quality of VI-RADS MRI examinations. Nevertheless, the American College of Radiology VI-RADS steering committee recently published the VI-RADS Quality Score which provides stratification of the levels of quality (inadequate, adequate, and optimal) using criteria based on signal-to-noise ratio, visualization of anatomic landmarks, and artifacts [32]. Table 2 summarizes some of the important elements within the VI-RADS Quality Score system, such as the minimum technical requirements and recommendations for each quality score category. While this is encouraging news, implementation of quality assurance/improvement efforts and future studies are being awaited that will test the VI-RADS Quality Score to see how it can impact clinical management as has been shown by the Prostate Imaging Quality scoring system for prostate MRI [33, 34]. For example, will the use of VI-RADS Quality Score improve the quality of bladder MRIs, increase the accuracy of determining MIBC, and ultimately improve the outcomes of patients with bladder cancer? A representative example of high-quality MRI showing exquisite morphological detail even in a subcentimeter tumor is shown in Fig. 4. Table 2Summary of VI-RADS Quality Score 1. Minimum technical requirements of each MRI sequence T2WIDWIDCE-MRI- Slice thickness ≤ 4 mm- ≥2 planes without fat saturation- Slice thickness ≤ 4 mm- High b-value (800–1000 s/mm2)- Slice thickness ≤ 3 mm- Fat suppression- 0.1 mmol/kg at 1.5–2.0 mL/sec- acquired 15 s after injection and then 4–6 acquisitions every 30 s 2. Specific elements that are assessed on each MRI sequence T2WIDWIDCE-MRI- Adequate SNR- Ability to clearly delineate the anatomy- Absence of significant artifacts- Ability to clearly delineate the anatomy- Absence of significant artifacts 3. Overall VI-RADS Quality Score and recommended action Inadequate (0) ≥ 2 MRI sequences are “inadequate”Repeat the scan if non-diagnostic examination Adequate (1) 2/3 MRI sequences are “adequate” or “optimal”Consider repeating the inadequate sequence if non-diagnostic examination Optimal (2) All 3 MRI sequences are “optimal”No action is needed Summary of VI-RADS Quality Score - Slice thickness ≤ 4 mm - ≥2 planes without fat saturation - Slice thickness ≤ 4 mm - High b-value (800–1000 s/mm2) - Slice thickness ≤ 3 mm - Fat suppression - 0.1 mmol/kg at 1.5–2.0 mL/sec - acquired 15 s after injection and then 4–6 acquisitions every 30 s - Adequate SNR - Ability to clearly delineate the anatomy - Absence of significant artifacts - Ability to clearly delineate the anatomy - Absence of significant artifacts Inadequate (0) ≥ 2 MRI sequences are “inadequate” Adequate (1) 2/3 MRI sequences are “adequate” or “optimal” Optimal (2) All 3 MRI sequences are “optimal” Bladder MRI of 68-year-old man demonstrating optimal quality allowing detailed morphological assessment of subcentimeter bladder tumor. (A) T2-weighted imaging, (B) diffusion-weighted imaging, (C) apparent diffusion coefficient map, and (D) dynamic contrast-enhanced MRI shows 0.7 cm polypoid tumor (arrow) at left lateral bladder wall which demonstrates restricted diffusion and avid enhancement. In contrast, the thickened inner layer (arrowhead) shows distinct imaging features from the tumor itself with (A) high T2 signal, (B-C) lack of restricted diffusion, and (D) minimal enhancement. Note that the quality of the multiparametric MRI is “optimal” as there is high signal-to-noise ratio, anatomical landmarks are clearly visualized, and there are no imaging artifacts. Transurethral resection revealed non-muscle-invasive bladder cancer “New game, new rules”: VI-RADS in the post-treatment setting While VI-RADS focuses on the pretreatment setting, bladder MRI has also demonstrated its strength in post-treatment assessment. Specifically, there has been continued interest and accumulating data showing that MRI can be used to assess for residual tumor after neoadjuvant therapies (including chemotherapy and immunotherapy) and stratify patients into different prognostic groups [35, 36], and, arguably, in the near future, select patients for bladder-sparing treatments [37]. MRI for local restaging after systematic therapy is especially attractive in the context of new and interesting data coming from recent trials. In the phase 3 KEYNOTE-B15 study, neoadjuvant and adjuvant enfortumab vedotin + pembrolizumab (antibody-drug conjugate and immunotherapy combination) for MIBC demonstrated superior oncologic outcomes compared with conventional neoadjuvant chemotherapy (gemcitabine + cisplatin); yet the pathological complete response rate at radical cystectomy was only 56% [38]. In addition, an integrated analysis of the phase II RETAIN 1 and 2 trials revealed that while post-treatment circulating tumor deoxyribonucleic acid (ctDNA) status was strongly predictive of metastatic control in patients with MIBC who underwent neoadjuvant therapy followed by cystectomy-sparing active surveillance, it was not reliably predictive of local-only recurrences–19 out of 22 of those with recurrences were ctDNA negative [39]. As such MRI may be able to provide complementary and synergistic information to liquid biopsy techniques to improve management of patients with MIBC. While MRI is recommended before and after systemic therapy or trimodal therapy based on expert consensus, challenges still remain in evaluating the presence and extent of residual tumor after such therapies due to confounding treatment-related changes (e.g., fibrosis, edema, or necrosis) [6]. To address this, an interpretation framework expanding on but based on the principles of VI-RADS called neoadjuvant chemotherapy VI-RADS (nacVI-RADS) was introduced in 2022 [40]. A few recent studies have validated nacVI-RADS for treatment response assessment after different systemic therapies (including conventional chemotherapy and immunotherapy) demonstrating its feasibility with sensitivities and specificities of 67–85%, 63–96% for detecting residual tumor [41, 42]. Figure 5 shows a representative example of a patient with MIBC in which nacVI-RADS is used for response assessment after chemotherapy. More research is needed to establish the optimal way to assess treatment response and answer questions such as whether one umbrella scoring system is sufficient or if different criteria are needed to assess response after different types of therapies (e.g., radiotherapy). In addition, what is the best way to interpret post-neoadjuvant therapy MRIs when there is no available pre-treatment MRI (Fig. 6)? While ideally, comparison between two timepoints (as required by nacVI-RADS) will provide the most comprehensive information, pragmatic considerations may be needed for real-life situations where often no pre-treatment MRI is available. Application of nacVI-RADS before and after neoadjuvant chemotherapy for muscle-invasive bladder cancer in 68-year-old man. (A-C) Pre-treatment MRI; (A) T2-weighted imaging, (B) diffusion-weighted imaging, and (C) dynamic contrast-enhanced MRI shows 4.6 × 1.3 cm bladder tumor (arrow) diffusely involving muscularis propria of the left and posterior bladder wall (VI-RADS 4). (D-F) Post-treatment MRI; After neoadjuvant chemotherapy, (D) T2-weighted imaging, (E) diffusion-weighted imaging, and (F) dynamic contrast enhancement imaging show decrease in size of tumor (arrow) to 3.4 × 1.2 cm but persistent muscle invasion (i.e., lack of downstaging). Therefore, response to neoadjuvant therapy is categorized as nacVI-RADS 5. Radical cystectomy revealed deep muscularis propria involvement and perivesical soft tissue infiltration (ypT3) Persistent muscle-invasive bladder cancer after neoadjuvant chemotherapy in 69-year-old man without pre-treatment MRI. (A) T2-weighted imaging, (B) diffusion-weighted imaging, (C) apparent diffusion coefficient map, and (D) dynamic contrast-enhanced MRI shows 3.9 × 2.7 cm mass (arrow) at left posterolateral bladder wall with diffuse invasion of muscularis propria and extension into perivesical fat (arrowheads). While no pre-treatment MRI was available, it is clear from this single timepoint imaging that there was a suboptimal response to neoadjuvant chemotherapy. Cystectomy revealed residual muscle invasive bladder cancer with perivesical soft tissue infiltration (ypT3) “Practice makes perfect”: wider adoption, training, and implementation While there is no doubt that VI-RADS has had a large impact on the practice of bladder cancer imaging, it is also clear that the vast majority of the application of VI-RADS is currently limited to highly specialized academic centers that have resources to do so, which includes but is not limited to availability of MRI scanners, experienced technologists, radiologists with expertise in using VI-RADS, referring clinicians committed to MRI staging and follow-up, and supportive reimbursement rules of the country or territory [43]. While various strategies will be needed to overcome these barriers and to accelerate widespread global adoption and implementation of VI-RADS, we could imagine that improved training opportunities are an essential component in the radiologists’ perspective. Studies have shown that the diagnostic performance of VI-RADS increases with experience and that this can be achieved with dedicated interactive training programs that are composed of didactic lectures and interactive case reviews [44, 45]. Increased exposure to VI-RADS can be facilitated by societal efforts, such as in-person hands-on workshops like the “How to interpret VI-RADS” by the European Society of Urogenital Radiology or even on-line virtual platforms like the “E-course: MRI in bladder cancer” by the EAU [46, 47]. Finally, integration of VI-RADS into multidisciplinary tumor board settings can increase the familiarity of VI-RADS in referring physicians and collectively improve experience and expertise across systems and institutions. “Man and machine”: Seeing beyond VI-RADS with artificial intelligence Advances in artificial intelligence (AI) or deep learning are expected to have a great impact on the role of MRI and VI-RADS in bladder cancer management. There already have been studies demonstrating that AI algorithms can help improve image quality using deep learning reconstructions, perform reliable tumor segmentation, assist in assigning VI-RADS scores which can potentially improve inter-reader agreement, and even provide expert-level diagnostic performance in determining muscle invasion of bladder cancer [48–51]. Despite these promising results, lack of validation and integration into our daily clinical practice (e.g., Picture Archiving and Communication System) continue to be significant hurdles for the widespread use of AI. The integration of AI into bladder MRI protocols warrants further investigation to determine how it can assist in tumor grading, molecular subtyping, and survival outcomes. Leveraging AI-enhanced VI-RADS and MRI assessment could significantly strengthen the role of imaging within the multimodal management of bladder cancer [52]. Conclusion VI-RADS has standardized the way we assess muscle invasion of bladder cancer on MRI. While it demonstrates good diagnostic performance and inter-reader agreement, a multifaceted approach to address unmet needs and pitfalls is needed prior to developing an updated version of VI-RADS. This includes questions with regards to omitting the use of contrast media, introducing quantitative metrics, consideration of anatomical locations prone to pitfalls, validating the new VI-RADS Quality Score, applications of MRI in the post-treatment setting, overcoming hurdles to improve training and widespread adoption, and integration of AI. Addressing these questions will bring us one step forward to a further improved VI-RADS–perhaps version 2.0–and ultimately improve outcomes of patients with bladder cancer. Abbreviations Apparent diffusion coefficient Artificial intelligence Biparametric magnetic resonance imaging Confidence interval Circuiting tumor deoxyribonucleic acid Dynamic contrast-enhanced Diffusion-weighted imaging European Association of Urology Muscle-invasive bladder cancer Multiparametric magnetic resonance imaging Neoadjuvant chemotherapy Vesical Imaging-Reporting and Data System Non-muscle-invasive bladder cancer Tumor-contact length Transurethral resection of bladder tumor Vesical Imaging-Reporting and Data System Author contributions S.W. prepared the original draft. All other authors reviewed and edited the manuscript. Funding No funding received. Data availability No datasets were generated or analysed during the current study. Declarations Ethical approval Not applicable in the review article. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement No datasets were generated or analysed during the current study.