Artificial intelligence for gastroenterology: Singapore artificial intelligence for Gastroenterology Working Group Position Statement

BACKGROUND

Successful implementation of artificial intelligence in gastroenterology and hepatology practice requires more than technology. There are ethical, legal, and social issues that need to be settled.

AIM

A group consisting of AI developers (engineer), AI users (gastroenterologist, hepatologist, and surgeon) and AI regulators (ethicist and administrator) formed a Working Group to draft these Positions Statements with the objective of arousing public and professional interest and dialogue, to promote ethical considerations when implementing AI technology, to suggest to policy makers and health authorities relevant factors to take into account when approving and regulating the use of AI tools, and to engage the profession in preparing for change in clinical practice.

STATEMENTS

These series of Position Statements point out the salient issues to maintain the trust between care provider and care receivers, and to legitimize the use of a non-human tool in healthcare delivery. It is based on fundamental principles such as respect, autonomy, privacy, responsibility, and justice. Enforcing the use of AI without considering these factor risk damaging the doctor-patient relationship.

The impact of lymph node ratio on disease recurrence in papillary thyroid microcarcinoma

INTRODUCTION

Lymph node (LN) metastases in papillary thyroid microcarcinomas (microPTCs) are common. The lymph node ratio (LNR) has been proposed as a risk factor for recurrence in papillary thyroid cancer. However, its relevance in microPTC is undetermined.

METHODS

Patients who underwent resection of their microPTC with concomitant LN clearance between 2005 and 2018 were identified. The LNR was calculated as the ratio of positive LNs to the total number of LNs.

RESULTS

Data on 50 patients (36 female [72%]; median age 47 years [range: 19-84]) who underwent LN clearance (28 central [56%] vs 22 central + lateral [44%]) were analysed. Positive LNs were found in over two-thirds of the patients (n = 34; 68%). After a median follow-up of 61 months, 14 patients (28%) had developed recurrence. Positive LNs were not found to impact recurrence-free survival; extranodal extension and an LNR ≥ 0.26 were found to significantly increase the risk of recurrence on unadjusted analyses (p < 0.05).

CONCLUSIONS

LN metastases are frequent among patients with microPTC. A higher LNR seems to be associated with recurrence. Additional studies are needed to further clarify these findings and to assess the possible role of LNR in treatment and surveillance.

Context-aware deep network for coronary artery stenosis classification in coronary CT angiography

Automatic coronary artery stenosis grading plays an important role in the diagnosis of coronary artery disease. Due to the difficulty of learning the informative features from varying grades of stenosis, it is still a challenging task to identify coronary artery stenosis from coronary CT angiography (CCTA). In this paper, we propose a context-aware deep network (CADN) for coronary artery stenosis classification. The proposed method integrates 3D CNN with Transformer to improve the feature representation of coronary artery stenosis in CCTA. We evaluate the proposed method on a multicenter dataset (APOLLO study with NCT05509010). Experimental results show that our proposed method can achieve the accuracy of 0.84, 0.83, and 0.86 for stenosis diagnosis on the lesion, artery, and patient levels, respectively.

Machine learning accurately quantifies epicardial adipose tissue from non-contrast CT images in coronary artery disease

Funding Acknowledgements

Type of funding sources: Other. Main funding source(s): Industry Alignment Fund – Pre-positioning Programme

Background

Epicardial adipose tissue (EAT) is the visceral fat deposit within the pericardium that surrounds the heart and the coronary arteries. EAT volume measured from non-contrast CT (NCCT) has been demonstrated to be significantly associated with adverse cardiovascular risk,1 particularly in patients with coronary artery disease.2 However, routine measurement of EAT volume is still challenging in clinical practice, as it is a tedious manual process and prone to human error.

Purpose

We aimed to develop a fully automated AI toolkit (i.e., AI EAT) for the quantification of EAT from routine NCCT scans and assess its performance in reference to clinical ground truth.

Methods

This is a multicenter study which performs CT scans in 5000 Asian Admixture patients (APOLLO study NCT05509010). In the current stage of this study, NCCT data analysis were conducted in 551 patients with 26,037 images. AI EAT was developed via a novel deep learning framework using an ensemble region-based UNet. The region-based UNet uses 2 component UNet models to perform segmentation of pericardium at the apex region and non-apex region (middle and basal). EAT volume was obtained by automated thresholding of the voxels (-190 to -30 Hounsfield Unit) within the pericardium (Figure 1). The network was trained in 501 patients with 23,712 NCCT images and tested in 50 patients with 2,325 NCCT images. The performance of AI EAT was evaluated with respect to clinical ground truth using Dice similarity coefficient (DSC), Pearson correlation, and Bland-Altman analysis.

Results

The AI EAT quantification process took less than 10 seconds per subject, compared with 20-30 minutes for expert readers. Compared to clinical ground truth, our AI EAT achieved a DSC of 0.96±0.01 and 0.91±0.02 for pericardium and EAT segmentations, respectively. There was strong agreement between the AI EAT and clinical ground truth in deriving the EAT volume (r=0.99, P&lt;0.001) with minimal error of 7±5%.

Conclusion

End-to-end deep learning system accurately quantifies epicardial adipose tissue in standard NCCT images without manual segmentation.

Health Care Analytics With Time-Invariant and Time-Variant Feature Importance to Predict Hospital-Acquired Acute Kidney Injury: Observational Longitudinal Study

Background

Acute kidney injury (AKI) develops in 4% of hospitalized patients and is a marker of clinical deterioration and nephrotoxicity. AKI onset is highly variable in hospitals, which makes it difficult to time biomarker assessment in all patients for preemptive care.

Objective

The study sought to apply machine learning techniques to electronic health records and predict hospital-acquired AKI by a 48-hour lead time, with the aim to create an AKI surveillance algorithm that is deployable in real time.

Methods

The data were sourced from 20,732 case admissions in 16,288 patients over 1 year in our institution. We enhanced the bidirectional recurrent neural network model with a novel time-invariant and time-variant aggregated module to capture important clinical features temporal to AKI in every patient. Time-series features included laboratory parameters that preceded a 48-hour prediction window before AKI onset; the latter’s corresponding reference was the final in-hospital serum creatinine performed in case admissions without AKI episodes.

Results

The cohort was of mean age 53 (SD 25) years, of whom 29%, 12%, 12%, and 53% had diabetes, ischemic heart disease, cancers, and baseline eGFR <90 mL/min/1.73 m², respectively. There were 911 AKI episodes in 869 patients. We derived and validated an algorithm in the testing dataset with an AUROC of 0.81 (0.78-0.85) for predicting AKI. At a 15% prediction threshold, our model generated 699 AKI alerts with 2 false positives for every true AKI and predicted 26% of AKIs. A lowered 5% prediction threshold improved the recall to 60% but generated 3746 AKI alerts with 6 false positives for every true AKI. Representative interpretation results produced by our model alluded to the top-ranked features that predicted AKI that could be categorized in association with sepsis, acute coronary syndrome, nephrotoxicity, or multiorgan injury, specific to every case at risk.

Conclusions

We generated an accurate algorithm from electronic health records through machine learning that predicted AKI by a lead time of at least 48 hours. The prediction threshold could be adjusted during deployment to optimize recall and minimize alert fatigue, while its precision could potentially be augmented by targeted AKI biomarker assessment in the high-risk cohort identified.

Preoperative scar perception study comparing 'scarless' in the neck endoscopic thyroidectomy with open thyroidectomy: a cross-sectional study

INTRODUCTION

Open thyroidectomy is the most common approach to thyroid surgery. However, 'scarless' (in the neck) endoscopic thyroidectomy, consisting of endoscopic and robotic surgery, is progressively being adopted for its perceived cosmetic benefits. This study aims to determine the patient's preferred surgical approach and to identify the factors that influence their decision.

MATERIALS AND METHODS

A pilot study consisting of 100 patients with a surgical thyroid disorder were prospectively recruited from a single tertiary centre. An interviewer-administered survey was conducted. Demographic, socioeconomic status, scar perception and an adapted body image scale were evaluated to identify factors that shaped the patient's perception of the surgical approach.

RESULTS

The mean age of participants was 54.5 ± 13.0 years; 72% were women and 87% Chinese. Of the 100 patients, 75 patients considered scarless endoscopic thyroidectomy as their preferred surgical approach while 25 patients opted for open thyroid surgery. Improvement in scar perception score between scarless endoscopic thyroidectomy and open thyroid surgery is associated with an increased willingness to choose scarless endoscopic thyroidectomy. The mean body image scale score was 6.9 ± 2.8, indicating no statistical difference between the surgical approaches. On multivariate analysis, improvement in scar perception score (odds ratio 3.38, 95% confidence interval 1.11-10.29) and having surgeon recommendation (odds ratio 6.38, 95% confidence interval 1.80-22.63) were independently associated with interest in scarless endoscopic thyroidectomy.

CONCLUSION

Patients interest in undergoing scarless endoscopic thyroidectomy is driven by improved scar perception and surgeon's recommendation compared with open thyroid surgery.

Binding of aberrant glycoproteins recognizable by Helix pomatia agglutinin in adrenal cancers

Background

Aberrant glycosylation is a hallmark of cancer cells and plays an important role in oncogenesis and cancer progression including metastasis. This study aimed to assess alteration in cellular glycosylation, detected by lectin Helix pomatia agglutinin (HPA) binding, in adrenal cancers and to determine whether such altered glycosylation has prognostic significance.

Methods

HPA binding lectin histochemistry was performed on archival paraffin wax‐embedded specimens of adrenocortical cancers excised from patients attending two tertiary referral centres. Benign tumours were used as controls. Demographic, histological and survival data were collected and compared between patients with HPA‐positive and HPA‐negative tumours.

Results

Thirty‐two patients were treated for adrenal cancer between 2000 and 2016; their median age was 49 (range 23–79) years. Fifteen patients had functioning tumours (14 adrenal Cushing's tumours and 1 Conn's tumour). Mean(s.d.) tumour size was 127·71(49·70) mm. None of 10 control tumours expressed HPA‐binding glycoproteins. Invasion was associated with HPA‐binding glycoproteins (P = 0·018). Local recurrence or metastatic disease did not significantly differ between HPA‐positive and HPA‐negative adrenocortical cancers. Overall survival was significantly longer in patients with HPA‐negative tumours (median survival not reached versus 22 months in patients with HPA‐positive tumours; P = 0·002).

Conclusion

Altered cellular glycosylation detected by lectin HPA is associated with poor survival in patients with adrenocortical cancer.

Pseudogout - a rare manifestation of hungry bone syndrome after focused parathyroidectomy

Pseudogout, also known as calcium pyrophosphate deposition disease, is a rheumatological condition arising from accumulation of calcium pyrophosphate dihydrate crystals in connective tissues. We present a case of a 56-year-old Bangladeshi woman who underwent focused right inferior parathyroidectomy for primary hyperparathyroidism from a right inferior parathyroid adenoma. On the first post-operative day, she complained of left elbow painful swelling with redness and warmth. Arthrocentesis of left elbow was done due to suspicion of septic arthritis. Two weeks prior to this surgery, she had sudden bilateral knee swelling was diagnosed in her home country of bilateral knee osteoarthritis with effusion and arthrocentesis showed no crystals. Aspiration of left elbow showed calcium pyrophosphate crystals, associated with post parathyroidectomy hypocalcemia, hypomagnesemia confirming pseudogout. Her uric acid level was normal. Bilateral wrist x-rays showed triangular fibrocartilage complex chondrocalcinosis. The patient's condition improved with colchicine and naproxen, as well as calcium and magnesium replacement. Her left elbow swelling and pain resolved. Pseudogout flare is a rare but known sequelae after parathyroidectomy. Early recognition and expeditious treatment is essential.

Rising incidence of thyroid cancer in Singapore not solely due to micropapillary subtype

Introduction The annual incidence of thyroid cancer is known to vary with geographic area, age and gender. The increasing incidence of thyroid cancer has been attributed to increase in detection of micropapillary subtype, among other factors. The aim of the study was to investigate time trends in the incidence of thyroid cancer in Singapore, an iodine-sufficient area. Materials and methods Data retrieved from the Singapore National Cancer Registry on all thyroid cancers that were diagnosed from 1974 to 2013 were reviewed. We studied the time trends of thyroid cancer based on gender, race, pathology and treatment modalities where available. Results The age-standardised incidence rate of thyroid cancer increased to 5.6/100,000 in 2013 from 2.5/100,000 in 1974. Thyroid cancer appeared to be more common in women, with a higher incidence in Chinese and Malays compared with Indians. Papillary carcinoma is the most common subtype. The percentage of papillary microcarcinoma has remained relatively stable at around 38% of all papillary cancers between 2007 and 2013. Although the incidence of thyroid cancer has increased since 1974, the mortality rate has remained stable. Conclusion This trend of increase in incidence of thyroid cancer in Singapore compares with other published series; however, the rise seen was not solely due to micropapillary type. Thyroid cancer was also more common in Chinese and Malays compared with Indians for reasons that needs to be studied further.

Mediastinal parathyroid adenomas and their surgical implications

Primary hyperparathyroidism is a relatively common problem encountered by any endocrine surgical unit. Ectopic parathyroid adenomas have been known to be a common cause of persistent hyperparathyroidism after surgery. A common site of the missed ectopic gland will be that in the mediastinum. However, with the increasing improvement in available imaging, it is likely that this can be diagnosed preoperatively. The surgical approach to the mediastinal parathyroid has also changed vastly over the last decade from maximally invasive to minimally invasive with minimal complications. We provide a review on the entity of mediastinal parathyroid adenomas and their surgical implications.