Development and validation of artificial neural networks model for detection of Barrett's neoplasia: a multicenter pragmatic nonrandomized trial (with video).

BACKGROUND AND AIMS: The aim of this study was to develop and externally validate a computer-aided detection (CAD) system for the detection and localization of Barrett's neoplasia and assess its performance compared with that of general endoscopists in a statistically powered multicenter study by using real-time video sequences. METHODS: In phase 1, the hybrid visual geometry group 16-SegNet model was trained by the use of 75,198 images and videos (96 patients) of neoplastic and 1,014,973 images and videos (65 patients) of nonneoplastic Barrett's esophagus. In phase 2, image-based validation was performed on a separate dataset of 107 images (20 patients) of neoplastic and 364 images (14 patients) of nonneoplastic Barrett's esophagus. In phase 3 (video-based external validation) we designed a real-time video-based study with 32 videos (32 patients) of neoplastic and 43 videos (43 patients) of nonneoplastic Barrett's esophagus from 4 European centers to compare the performance of the CAD model with that of 6 nonexpert endoscopists. The primary endpoint was the sensitivity of CAD diagnosis of Barrett's neoplasia. RESULTS: In phase 2, CAD detected Barrett's neoplasia with sensitivity, specificity, and accuracy of 95.3%, 94.5%, and 94.7%, respectively. In phase 3, the CAD system detected Barrett's neoplasia with sensitivity, specificity, negative predictive value, and accuracy of 93.8%, 90.7%, 95.1%, and 92.0%, respectively, compared with the endoscopists' performance of 63.5%, 77.9%, 74.2%, and 71.8%, respectively (P < .05 in all parameters). The CAD system localized neoplastic lesions with accuracy, mean precision, and mean intersection over union of 100%, 0.62, and 0.54, respectively, when compared with at least 1 of the expert markings. The processing speed of the CAD detection and localization were 5 ms/image and 33 ms/image, respectively. CONCLUSION: To our knowledge, this is the first study describing external (multicenter) validation of AI algorithms for the detection of Barrett's neoplasia on real-time endoscopic videos. The CAD system in this study significantly outperformed nonexpert endoscopists on real-time video-based assessment, achieving >90% sensitivity for neoplasia detection. This result needs to be validated during real-time endoscopic assessment.

Randomized controlled trial of EndoRings assisted colonoscopy versus standard colonoscopy.

OBJECTIVES: The EndoRings device is a distal attachment consisting of two layers of circular flexible rings that evert mucosal folds. The aim of this study was to investigate whether EndoRing assisted colonoscopy (ER) improves polyp and adenoma detection compared to standard colonoscopy (SC). METHODS: Multicenter, parallel group, randomized controlled trial. RESULTS: Total of 556 patients randomized to ER (n = 275) or SC (n = 281). Colonoscopy completed in 532/556 (96%) cases. EndoRings removed in 74/275 (27%) patients. Total number of polyps in ER limb 582 vs. 515 in SC limb, P = 0.04. Total number of adenomas in ER limb 361 vs. 343 for SC limb, P = 0.49. A statistically significant difference in the mean number of polyps per patient in both the intention to treat (1.84 SC vs. 2.10 ER, P = 0.027) and per protocol (PP) (1.84 SC vs. 2.25 ER, P = 0.004). CONCLUSIONS: Our study shows promise for the EndoRings device to improve polyp detection.

A brave new world: the gastrointestinal lab during and after a pandemic.

PURPOSE OF REVIEW: The COVID-19 global pandemic resulted in a radical change in the provision and delivery of endoscopy services worldwide. As we emerge from this pandemic, various strategies were advocated to resume endoscopy whilst prioritizing the safety and wellbeing of patients and staff. RECENT FINDINGS: This review summarizes the main changes including infection control and prevention measures in endoscopy and explores the overarching impact of the pandemic on the gastrointestinal lab. Various solutions are outlined to enable the well tolerated resumption of endoscopy services including retention of certain infection control measures, use of personal protective equipment, testing and vaccination. Strategies to deal with the mounting backlog of cases are also discussed. SUMMARY: The COVID-19 pandemic wreaked havoc on healthcare systems worldwide and affected the provision and delivery of gastrointestinal diagnostic services, such as endoscopy necessitating a new way of working and an emphasis on infection control and better use of technology that are likely to be here to stay in the post pandemic era.

Multicentre randomised controlled trial on virtual chromoendoscopy in the detection of neoplasia during colitis surveillance high-definition colonoscopy (the VIRTUOSO trial).

BACKGROUND: Longstanding colonic IBD increases the risk of developing colorectal cancer. The utility of chromoendoscopy with standard-definition white light technology has been established. However, the use of high-definition virtual chromoendoscopy (HDV) in colitis surveillance remains undefined. OBJECTIVE: To compare the performance of HDV (i-scan OE mode 2) with high-definition white light (HDWL) for detection of neoplasia in patients with IBD undergoing surveillance colonoscopy. Additionally, we assessed the utility of protocol-guided quadrantic non-targeted biopsies. DESIGN: A multioperator randomised controlled trial was carried out in two centres in the UK. Total of 188 patients (101 men, mean age 54) with longstanding ulcerative or Crohn's colitis were randomised, prior to starting the surveillance colonoscopy, to using either HDV (n=94) or HDWL (n=94) on withdrawal. Targeted and quadrantic non-targeted biopsies were taken in both arms per-randomisation protocol. The primary outcome was the difference in neoplasia detection rate (NDR) between HDV and HDWL. RESULTS: There was no significant difference between HDWL and HDV for neoplasia detection. The NDR was not significantly different for HDWL (24.2%) and HDV (14.9%) (p=0.14). All intraepithelial neoplasia (IEN) detected contained low-grade dysplasia only. A total of 6751 non-targeted biopsies detected one IEN only. The withdrawal time was similar in both arms of the study; median of 24 min (HDWL) versus 25.5 min (HDV). CONCLUSION: HDV and HDWL did not differ significantly in the detection of neoplasia. Almost all neoplasia were detected on targeted biopsy or resection. Quadrantic non-targeted biopsies have negligible additional gain. TRIAL REGISTRATION NUMBER: Clinical Trial.gov ID NCT02822352.

A novel self-assembling peptide for hemostasis during endoscopic submucosal dissection: a randomized controlled trial.

BACKGROUND: Endoscopic submucosal dissection (ESD) is associated with a risk of bleeding. Bleeding is usually treated with diathermy, although this does carry a risk of mucosal thermal injury. Purastat is a topical hemostat that may be effective in controlling bleeding during ESD, thereby reducing the use of heat therapy. The aim of this study was to assess the reduction in heat therapy used in the interventional group (Purastat) compared with the control group. The secondary aims were to compare the procedure length, time for hemostasis, delayed bleeding rate, adverse events, and wound healing between the groups. METHODS: This was a single-center randomized controlled trial of 101 patients undergoing ESD. Participants were randomized to a control group where diathermy was used to control bleeding or an interventional group where Purastat could be used. Follow-up endoscopy was performed at 4 weeks to assess wound healing. RESULTS: There was a significant reduction in the use of heat therapy for intraprocedural hemostasis in the interventional group compared with controls (49.3 % vs. 99.6 %, P < 0.001). There were no significant differences in the procedure length, time for hemostasis, and delayed bleeding rate between the groups. Complete wound healing at 4 weeks was noted in 48.8 % of patients in the interventional group compared with 25.0 % of controls (P = 0.02). CONCLUSIONS: This study has demonstrated that Purastat is an effective hemostat that can reduce the need for heat therapy for bleeding during ESD. It may also have a role in improving post-resection wound healing.

Optical diagnosis of colorectal polyp images using a newly developed computer-aided diagnosis system (CADx) compared with intuitive optical diagnosis.

BACKGROUND: Optical diagnosis of colorectal polyps remains challenging. Image-enhancement techniques such as narrow-band imaging and blue-light imaging (BLI) can improve optical diagnosis. We developed and prospectively validated a computer-aided diagnosis system (CADx) using high-definition white-light (HDWL) and BLI images, and compared the system with the optical diagnosis of expert and novice endoscopists. METHODS: CADx characterized colorectal polyps by exploiting artificial neural networks. Six experts and 13 novices optically diagnosed 60 colorectal polyps based on intuition. After 4 weeks, the same set of images was permuted and optically diagnosed using the BLI Adenoma Serrated International Classification (BASIC). RESULTS: CADx had a diagnostic accuracy of 88.3 % using HDWL images and 86.7 % using BLI images. The overall diagnostic accuracy combining HDWL and BLI (multimodal imaging) was 95.0 %, which was significantly higher than that of experts (81.7 %, P = 0.03) and novices (66.7 %, P < 0.001). Sensitivity was also higher for CADx (95.6 % vs. 61.1 % and 55.4 %), whereas specificity was higher for experts compared with CADx and novices (95.6 % vs. 93.3 % and 93.2 %). For endoscopists, diagnostic accuracy did not increase when using BASIC, either for experts (intuition 79.5 % vs. BASIC 81.7 %, P = 0.14) or for novices (intuition 66.7 % vs. BASIC 66.5 %, P = 0.95). CONCLUSION: CADx had a significantly higher diagnostic accuracy than experts and novices for the optical diagnosis of colorectal polyps. Multimodal imaging, incorporating both HDWL and BLI, improved the diagnostic accuracy of CADx. BASIC did not increase the diagnostic accuracy of endoscopists compared with intuitive optical diagnosis.

Recovery of endoscopy services in the era of COVID-19: recommendations from an international Delphi consensus.

The COVID-19 pandemic has had a profound impact on provision of endoscopy services globally as staff and real estate were repurposed. As we begin to recover from the pandemic, a cohesive international approach is needed, and guidance on how to resume endoscopy services safely to avoid unintended harm from diagnostic delays. The aim of these guidelines is to provide consensus recommendations that clinicians can use to facilitate the swift and safe resumption of endoscopy services. An evidence-based literature review was carried out on the various strategies used globally to manage endoscopy during the COVID-19 pandemic and control infection. A modified Delphi process involving international endoscopy experts was used to agree on the consensus statements. A threshold of 80% agreement was used to establish consensus for each statement. 27 of 30 statements achieved consensus after two rounds of voting by 34 experts. The statements were categorised as pre-endoscopy, during endoscopy and postendoscopy addressing relevant areas of practice, such as screening, personal protective equipment, appropriate environments for endoscopy and infection control precautions, particularly in areas of high disease prevalence. Recommendations for testing of patients and for healthcare workers, appropriate locations of donning and doffing areas and social distancing measures before endoscopy are unique and not dealt with by any other guidelines. This international consensus using a modified Delphi method to produce a series of best practice recommendations to aid the safe resumption of endoscopy services globally in the era of COVID-19.

Development and validation of the international Blue Light Imaging for Barrett's Neoplasia Classification.

BACKGROUND AND AIMS: Detecting subtle Barrett's neoplasia during surveillance endoscopy can be challenging. Blue-light imaging (BLI) is a novel advanced endoscopic technology with high-intensity contrast imaging that may improve the identification of Barrett's neoplasia. The aim of this study was to develop and validate the first classification to enable characterization of neoplastic and non-neoplastic Barrett's esophagus using BLI. METHODS: In phase 1, descriptors pertaining to neoplastic and non-neoplastic Barrett's esophagus were identified to form the classification, named the Blue Light Imaging for Barrett's Neoplasia Classification (BLINC). Phase 2 involved validation of these component criteria by 10 expert endoscopists assessing 50 BLI images. In phase 3, a web-based training module was developed to enable 15 general (nonexpert) endoscopists to use BLINC. They then validated the classification with an image assessment exercise in phase 4, and their pre- and post-training results were compared. RESULTS: In phase 1 the descriptors were grouped into color, pit, and vessel pattern categories to form the classification. In phase 2 the sensitivity of neoplasia identification was 96.0% with a very good level of agreement among the experts (κ = .83). In phase 3, 15 general endoscopists completed the training module. In phase 4 their pretraining sensitivity (85.3%) improved significantly to 95.7% post-training with a good level of agreement (κ = .67). CONCLUSIONS: We developed and validated a new classification system (BLINC) for the optical diagnosis of Barrett's neoplasia using BLI. Despite the limitations of this image-based study with a high prevalence of neoplasia, we believe it has the potential to improve the optical diagnosis of Barrett's neoplasia given the high degree of sensitivity (96%) noted. It is also a promising tool for training in Barrett's esophagus optical diagnosis using BLI.

Outcomes from an international multicenter registry of patients with acute gastrointestinal bleeding undergoing endoscopic treatment with Hemospray.

BACKGROUND AND AIM: Acute gastrointestinal bleeding carries poor outcomes unless prompt endoscopic hemostasis is achieved. Mortality in these patients remains significant. Hemospray is a novel intervention that creates a mechanical barrier over bleeding sites. We report the largest dataset of patient outcomes after treatment with Hemospray from an international multicenter registry. PATIENTS AND METHODS: Prospective data (Jan 2016-May 2018) from 12 centers across Europe were collected. Immediate hemostasis was defined as endoscopic cessation of bleeding within 5 min after application of Hemospray. Rebleeding was defined as subsequent drop in hemoglobin, hematemesis, persistent melena with hemodynamic compromise post-therapy. RESULTS: Three hundred and fourteen cases were recruited worldwide (231 males, 83 females). Median pretreatment Blatchford score was 11 (IQR: 8-14) and median complete Rockall score (RS) was 7 (IQR: 6-8) for all patients. Peptic ulcer disease (PUD) was the most common pathology (167/314 = 53%) and Forrest Ib the most common bleed type in PUD (100/167 = 60%). 281 patients (89.5%) achieved immediate hemostasis after successful endoscopic therapy with Hemospray. Rebleeding occurred in 29 (10.3%) of the 281 patients who achieved immediate hemostasis. Seven-day and 30-day all-cause mortality were 11.5% (36/314) and 20.1% (63/314), respectively (lower than the predicted rates as per the RS). Similar hemostasis rates were noted in the Hemospray monotherapy (92.4%), combination therapy (88.7%) and rescue therapy (85.5%) groups. CONCLUSIONS: These data show high rates of immediate hemostasis overall and in all subgroups. Rebleeding and mortality rates were in keeping/lower than predicted rates.

International development and validation of a classification system for the identification of Barrett's neoplasia using acetic acid chromoendoscopy: the Portsmouth acetic acid classification (PREDICT).

BACKGROUND: Barrett's oesophagus is an established risk factor for developing oesophageal adenocarcinoma. However, Barrett's neoplasia can be subtle and difficult to identify. Acetic acid chromoendoscopy (AAC) is a simple technique that has been demonstrated to highlight neoplastic areas but lesion recognition with AAC remains a challenge, thereby hampering its widespread use. OBJECTIVE: To develop and validate a simple classification system to identify Barrett's neoplasia using AAC. DESIGN: The study was conducted in four phases: phase 1-development of component descriptive criteria; phase 2-development of a classification system; phase 3-validation of the classification system by endoscopists; and phase 4-validation of the classification system by non-endoscopists. RESULTS: Phases 1 and 2 led to the development of a simplified AAC classification system based on two criteria: focal loss of acetowhitening and surface patterns of Barrett's mucosa. In phase 3, the application of PREDICT (Portsmouth acetic acid classification) by endoscopists improved the sensitivity and negative predictive value (NPV) from 79.3% and 80.2% to 98.1% and 97.4%, respectively (p<0.001). In phase 4, the application of PREDICT by non-endoscopists improved the sensitivity and NPV from 69.6% and 75.5% to 95.9% and 96.0%, respectively (p<0.001). CONCLUSION: We developed and validated a classification system known as PREDICT for the diagnosis of Barrett's neoplasia using AAC. The improvement seen in the sensitivity and NPV for detection of Barrett's neoplasia in phase 3 demonstrates the clinical value of PREDICT and the similar improvement seen among non-endoscopists demonstrates the potential for generalisation of PREDICT once proven in real time.

Complex early Barrett's neoplasia at 3 Western centers: European Barrett's Endoscopic Submucosal Dissection Trial (E-BEST).

BACKGROUND AND AIMS: Endoscopic submucosal dissection (ESD) is an effective technique to resect early Barrett's neoplasia and has advantages over conventional EMR in that it enables en-bloc resection and accurate histopathologic analysis of cancer resection margins. However, its long learning curve and higher adverse event rate have tempered its uptake in the West. We aimed to analyze the safety and efficacy of ESD when used to resect complex Barrett's neoplasia. The primary endpoint was the en-bloc and R0 resection rate. METHODS: This was a retrospective analysis of 143 ESDs for Barrett's neoplasia performed in 3 tertiary referral centers from 2008 to 2016. RESULTS: The mean lesion size was 31.1 mm (range, 5-90) and median follow-up time 21.6 months (interquartile range, 11.0-32.6). In total, 24.5% of lesions (35/143) were scarred after previous endoscopic resection, surgery, or radiotherapy. The en-bloc resection rate was 90.8% and R0 resection rate 79% in this series. The overall adverse event rate was 3.5% (1.4% bleeding, 0% perforation, and 2.1% stricture formation). The expanded curative resection rate was 65.8%, reflecting the R0 resection rate and proportion of cases with more advanced disease. Submucosal cancer was identified as a significant factor affecting the R0 resection rate. CONCLUSION: We demonstrated the feasibility and safety of ESD in the West for resection of complex Barrett's neoplasia including large, nodular, or scarred lesions. This is a safe and effective technique with a low adverse event rate when performed by an experienced operator. The en-bloc resection rate reached a plateau once 30 procedures had been performed.

International Delphi Consensus Study on disposable single-use endoscopy: A path to clinical adoption.

BACKGROUND/OBJECTIVE: Increasing infectious rate estimates and low microbiological surveillance affect safety of gastrointestinal endoscopy globally. Single use endoscopes and accessories have been claimed to improve safety, but there is lack of data on their indication and sustainability. We aimed to identify a series of best practice recommendations for the use of single use endoscopes and accessories using a modified Delphi. METHODS/DESIGN: Consensus statements for the use of single use endoscopy and accessories were developed using a modified Delphi process, utilizing an international endoscopist expert panel of 62 experts from 33 nations. The main steps in the process were selecting the consensus group, conducting systematic literature reviews, developing statements, and anonymous voting on the statements until consensus was reached. High-risk patients were defined as those with multi-drug-resistant infections, immunosuppressive medication or chemotherapy, post-transplantation, or with severe neutropenia. RESULTS: Of the 26 statements that were voted upon through two rounds, 17 statements reached consensus. Category 1: single use accessories (8 statements), related to defining recommendations for the use of single use accessories in all patient populations or high-risk patients. Category 2: clinical indication for single use endoscopes (9 statements), including indications to high-risk patients, protecting the endoscope apparatus and contamination measures in endoscopy units. Category 3: technical factors (4 statements), related to superior performance and technical specifications with the new innovation. Category 4: environmental issues (2 statements), concerning mechanisms that reduce the detrimental burden to the environment. Category 5: financial implications (3 statements), related to healthcare policies, cost neutrality and other financial associations of single use endoscopy. CONCLUSIONS: This is the first international initiative in determining clinical indications for single use endoscopy and accessories. The study's findings should serve as a framework for future physicians to guide future research and aid the proper evidence-based indications for the implementation of single use endoscopes in clinical practice.

UK endoscopy trainer survey: perspectives on current endoscopy training delivery, experience, barriers and opportunities.

Objective: UK endoscopy training is delivered by trainers possessing well developed endoscopy and teaching skills to help learners perform high-quality endoscopy. Train The Trainer (TTT) courses are effective, but additional trainer support is variable with little formal quality assurance. We performed a survey to map UK endoscopy training, assess trainer perspectives on training delivery and identify factors that would enhance training. Design/Method: An online survey was designed by trainer representatives, in collaboration with the JAG training committee, and collected responses from trainers registered on JAG endoscopy training system e-portfolio from April to June 2022. Results: There were 1024 responses from all trainer disciplines, with 813 (79%) completing TTT courses and 584 (57%) having job planned dedicated training lists (DTLs). Clinical endoscopists most frequently had job-planned DTLs (71%), and DTLs occurring at least weekly (58%). 293 (29%) respondents participated as course faculty. Trainers reported high levels of pre-procedure preparation, effective dialogue and frequent feedback. The DOPS forms were 'always/often' completed by 81% of clinical endoscopists, 73% of gastroenterologist and 58% of surgeons. 435 (42%) trainers never had peer feedback. Responses suggested training could improve by protecting training time, attending courses, participating as faculty and receiving feedback from experienced trainers. Conclusion: This survey demonstrates substantial proportions of highly motivated UK trainers who value time spent teaching and learning how to teach. Skills taught on the TTT courses are often actively used in everyday training. Improved trainer course access, protected training time and formal use of existing feedback tools by peers were highlighted as measures that could support trainers' development.

Erratum: Automatic textual description of colorectal polyp features: explainable artificial intelligence.

[This corrects the article DOI: 10.1055/a-2071-6652.].

Automatic textual description of colorectal polyp features: explainable artificial intelligence.

Computer-aided diagnosis systems (CADx) can improve colorectal polyp (CRP) optical diagnosis. For integration into clinical practice, better understanding of artificial intelligence (AI) by endoscopists is needed. We aimed to develop an explainable AI CADx capable of automatically generating textual descriptions of CRPs. For training and testing of this CADx, textual descriptions of CRP size and features according to the Blue Light Imaging (BLI) Adenoma Serrated International Classification (BASIC) were used, describing CRP surface, pit pattern, and vessels. CADx was tested using BLI images of 55 CRPs. Reference descriptions with agreement by at least five out of six expert endoscopists were used as gold standard. CADx performance was analyzed by calculating agreement between the CADx generated descriptions and reference descriptions. CADx development for automatic textual description of CRP features succeeded. Gwet's AC1 values comparing the reference and generated descriptions per CRP feature were: size 0.496, surface-mucus 0.930, surface-regularity 0.926, surface-depression 0.940, pits-features 0.921, pits-type 0.957, pits-distribution 0.167, and vessels 0.778. CADx performance differed per CRP feature and was particularly high for surface descriptors while size and pits-distribution description need improvement. Explainable AI can help comprehend reasoning behind CADx diagnoses and therefore facilitate integration into clinical practice and increase trust in AI.

Endoscopy training in the UK pre-COVID-19 environment: a multidisciplinary survey of endoscopy training and the experience of reciprocal feedback.

OBJECTIVE: Training in gastrointestinal endoscopy in the UK occurs predominantly in a real world one-to-one trainer to trainee interaction. Previous surveys have shown surgical and gastroenterology trainees have had mixed experiences of supervision and training, and no surveys have explored specifically the role of trainee to trainer feedback. This study aimed to explore the experience of training and of providing trainer feedback for all disciplines of endoscopy trainees. DESIGN/METHOD: An online survey designed in collaboration with Joint Advisory Committee training committee and trainee representatives was distributed from January 2020 but was interrupted by the COVID-19 pandemic and hence terminated early. RESULTS: There were 129 responses, including trainees from all disciplines and regions, of which 86/129 (66.7%) rated the culture in their endoscopy units favourably-either good or excellent. 65/129 (50.4%) trainees reported having one or more training lists allocated per week, with 41/129 (31.8%) reporting only ad hoc lists. 100/129 (77.5%) respondents were given feedback and 97/129 (75.2%) were provided with learning points from the list. 65/129 (50.4%) respondents reported their trainer completed a direct observation of procedure or direct observation of polypectomies. 73/129 (56.6%) respondents reported that they felt able to give feedback to their trainer, with 88/129 (68.2%) feeling they could do this accurately. Barriers to trainer feedback cited included time constraints, lack of anonymity and concerns about affecting the trainer-trainee relationship. CONCLUSION: Overall, the training environment has improved since previous surveys. There are still issues around interdisciplinary differences with some surgical trainees finding the training environment less welcoming, and trainee perceptions of hierarchical barriers and trainer responsiveness to feedback limiting the accuracy of their feedback.

Automatic image and text-based description for colorectal polyps using BASIC classification.

Colorectal polyps (CRP) are precursor lesions of colorectal cancer (CRC). Correct identification of CRPs during in-vivo colonoscopy is supported by the endoscopist's expertise and medical classification models. A recent developed classification model is the Blue light imaging Adenoma Serrated International Classification (BASIC) which describes the differences between non-neoplastic and neoplastic lesions acquired with blue light imaging (BLI). Computer-aided detection (CADe) and diagnosis (CADx) systems are efficient at visually assisting with medical decisions but fall short at translating decisions into relevant clinical information. The communication between machine and medical expert is of crucial importance to improve diagnosis of CRP during in-vivo procedures. In this work, the combination of a polyp image classification model and a language model is proposed to develop a CADx system that automatically generates text comparable to the human language employed by endoscopists. The developed system generates equivalent sentences as the human-reference and describes CRP images acquired with white light (WL), blue light imaging (BLI) and linked color imaging (LCI). An image feature encoder and a BERT module are employed to build the AI model and an external test set is used to evaluate the results and compute the linguistic metrics. The experimental results show the construction of complete sentences with an established metric scores of BLEU-1 = 0.67, ROUGE-L = 0.83 and METEOR = 0.50. The developed CADx system for automatic CRP image captioning facilitates future advances towards automatic reporting and may help reduce time-consuming histology assessment.