EUS in children with eosinophilic oesophagitis – a new method of measuring oesophageal total wall thickness area. An artificial intelligence application feasibility study. A pilot study
Slawomir Wozniak
Orcid profile, Radoslaw Kempinski
Orcid profile, Katarzyna Akutko
Orcid profile, Tomasz Pytrus
Orcid profile and Urszula Zaleska-Dorobisz
Orcid profile 
Article Category: Research paper
Published Online: Nov 08, 2024
Page range: 1 - 6
Received: Sep 25, 2023
Accepted: Jan 17, 2024
DOI: https://doi.org/10.15557/jou.2024.0020
Keywords
© 2024 Slawomir Wozniak et al., published by Sciendo
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
Eosinophilic oesophagitis (EoE) is diagnosed worldwide, and susceptibility to the disorder has increased over the last 10 years(1,2). Both children and adults are affected. In Poland, approximately 1 out of 5,000–10,000 individuals suffer from the condition, which gives a figure of approx. 3.8–7.6 thousand all together in Poland. Other authors estimate the incidence and prevalence as 10–57/100,000(3,4). The disease is more often diagnosed in North America, Europe or Australia than on other continents. Consequently, one can extrapolate the number of affected people across the world as circa 800,000–4,560,000. The majority of them are young (20–30 years old) male individuals, and every fifth diagnosis is established in childhood(5).
The number of diagnosed children is growing rapidly(2). The disease usually occurs in patients with a food and/or inhaled allergy(6,7) with age-related and atypical symptoms. No singular symptom is characteristic for EoE(2,6,8). No macroscopic changes found in the oesophagus are pathognomic for EoE(7). Extraction of specimens are necessary for establishing the diagnosis. Histologic findings of ≥15 eosinophils per high-power field serve as the diagnostic hallmark(4).
There are no straightforward correlations between patients’ symptoms, endoscopic changes, and histologic examination(8). The wall may be thicker in some diseases, among them in EoE. This abnormality can be diagnosed or traced with high-resolution endoscopic sonography (EUS). The oesophageal wall thickness (TWT) or chosen wall layer thickness can be assessed and measured(8–10). No Artificial Intelligence (AI) methods have been applied in diagnosing or treating EoE children so far(11,12).
Our objective was to introduce a formula for measuring the oesophageal total wall thickness area (TWTa) in oesophagitis which can be used as a foundation for developing an AI-based algorithm. The formula can also be applied in patients with other oesophageal diseases – for example scleroderma, oesophageal burns, surgical reconstructions and others.
This feasibility study was performed without the participation of patients – we analysed the posssibilties of applying a new method. The study was approved by the Institutional Review Board (ZAPUMW 29. 03. 2023/1). The IRB had waived the requirement to obtain informed consent.
The basic data for our calculations were taken from papers published by Dalby
During EUS, the circular sonographic scan of the oesophageal wall could be taken in the thoracic part of the oesophagus at the level of the Th3-Th7 vertebrae. The lumen and wall could be observed and analysed (Fig. 1).
In this pilot study, we did not perform any statistical analysis. The reason for this was the study design. We introduced a new method of measuring TWTa in patients with EoE.
When dilated, the oesophagus looks like a tube. When the US transducer is situated inside the oesophagus, its lumen is dilated, as shown in Fig. 2.
The oesophageal lumen area can be calculated according to the following formula describing the square area of the circle as ‘multiplication n by square radius of the circle’:
Where: AL – oesophageal lumen area; π = 3.14; r – oesophageal lumen radius.
The area of the oesophageal lumen plus TWT can be calculated according the same formula. The calculated value is the product of π and the square sum of adding the oesophageal lumen radius and TWT.
Where: Aw = the surface area of the circle composed of the oesophageal lumen and TWT; π = 3.14; Z = the radius of the lumen plus TWT.
The measurements may be simplified by using the formula for the area of the circular ring A = π (B2–b2); the product of n and subtraction square b (smaller radius) and square B (bigger radius).
The final formula for the oesophagus measurements can be shown as:
Where: ATWTa – the surface area of TWT; π = 3.14; Z – the sum of lumen radius and TWT; r – oesophageal lumen radius.
The radius should be measured three times in four different points, and the mean should be calculated. The measurement points should be determined according to the clock rule: at 1, 3, 6 and 9, and here established. The measurement positions are shown as * in Fig. 3.
Loff
Square area of the oesophageal lumen calculated according the formula πr2 (in mm2)
| Children’s age (weight according to our experience) | Square area of the oesophageal lumen (A = πr2; multiplication) in mm2 | ||
|---|---|---|---|
| Oesophagus at the 3rd Th vertebral level | Oesophagus at the 7th Th vertebral level | ||
| 0–3 years (3–18 kg) | Calculations for 3 kg | 3.14 and 3.5*2= |
3.14 and 4.42= |
| Calculations for 18 kg | 3.14 and 4.552= |
3,14 and 5.452 = |
|
| Above 3–6 years (13–29 kg) | Calculations for 13 kg | 3.14 and 4.22 = |
3.14 and 5.12= |
| Calculations for 29 kg | 3.14 and 5.352 = |
3.14 and 6.252= |
|
| Above 6–13 years (18–66 kg) | Calculations for 18 kg | 3.14 and 4.552= |
3.14 and 5.452 = |
| Calculations for 66 kg | 3.14 and 7.92= |
3.14 and 8.92= |
|
| Above 13 years (39–70 kg) | Calculations for 39 kg | 3.14 and 6.052= |
3.14 and 6.952= |
| Calculations for 70 kg | 3.14 and 8.22= |
3.14 and 9.22= |
|
The normal oesophageal lumen according to Loff
Dalby
Relying on these data (TWT measurements), we calculated the square areas of the thoracic part of the oesophagus according to the formulas described above. Our TWTa calculations are presented in Tab. 2.
Square area of the oesophageal lumen with TWT (in mm2)
| Children’s age (weight according to our experience) | Square area of the oesophageal lumen with TWT (A = nr2; multiplication) in mm2 | ||
|---|---|---|---|
| Oesophagus upper level | Oesophagus lower level | ||
| 0–3 years (3–18 kg) | Calculations for 3 kg | 3.14 and (3.5* + 4.2**)2= |
3.14 and (4.4 + 4.2)2= |
| Calculations for 18 kg | 3.14 and (4.55 + 4.2)2= |
3.14 and (5.45 + 4.2)2= |
|
| Above 3–6 years (13–29 kg) | Calculations for 13 kg | 3.14 and (4.2 + 4.2)2= |
3.14 and (5.1 + 4.2)2= |
| Calculations for 29 kg | 3.14 and (5.35 + 4.2)2= |
3.14 and (6.25 + 4.2)2= |
|
| Above 6–13 years (18–66 kg) | Calculations for 18 kg | 3.14 and (4.55 + 4.2)2= |
3.14 and (5.45 + 4.2)2= |
| Calculations for 66 kg | 3.14 and (7.9 + 4.2)2 = |
3.14 and (8.9 + 4.2)2= |
|
| Above 13 years (39–70 kg) | Calculations for 39 kg | 3.14 and (6.05 + 4.2)2= |
3.14 and (6.95 + 4.2)2= |
| Calculations for 70 kg | 3.14 and (8.2 + 4.2)2 = |
3.14 and (9.2 + 4.2)2= |
|
The normal oesophageal lumen according to Loff
The predicted TWTa values (in mm2) according to age are shown in Tab. 3.
TWTa calculated (mm2) at Th3-Th7 according to children’s age
| Children’s age (weight) | Square area of the TWT ring area (in mm2) | ||
|---|---|---|---|
| Oesophagus at the 3rd Th vertebral level | Oesophagus at the 7th Th vertebral level | ||
| 0–3 years (3–18 kg) | Calculations for 3 kg | ||
| Calculations for 18 kg | |||
| Above 3–6 years (13–29 kg) | Calculations for 13 kg | ||
| Calculations for 29 kg | |||
| Above 6–13 years (18–66 kg) | Calculations for 18 kg | ||
| Calculations for 66 kg | |||
| Above 13 years (39–70 kg) | Calculations for 39 kg | ||
| Calculations for 70 kg | |||
Explanation: how we calculated these values: * 147.71. Subtraction: 186.17 – 38.46 = 147.71 mm2; where 186.71 is the square area of oesophageal lumen and TWT calculated together; 38.46 = square area of the lumen
Predicted TWTa (in mm2) as min and max values for children of different weight categories are shown in Fig. 4. The curve is detailed to allow the sonographer to find the correct value at a glance. The calculated TWTa in young patients should be placed between the max and min values to be assessed as normal.
When TWTa is below 175.39, it shows that the oesophageal wall is too thin, and when it is above 199.13, the oesophageal wall is too thick, and the children should be diagnosed for EoE or other diseases in which the wall may be thickened.
EUS in children is mostly performed to diagnose choledocholithia-sis, pancreatic fluid collections, pancreatitis or pancreatic mass(14), so EoE is not the most common cause for performing the examination, but its usefulness has gradually increased(8,15). EUS in children should be performed by paediatric sonographers because of the difference in patients’ body size and the need for paediatric anaesthesia(16).
We introduced a new method for tracing fibrosis with oesophageal wall thickening risk, which increases circa 5% a year since establishing the diagnosis. If the disease duration is longer or patients do not respond to treatment, stenoses occur due to chronic inflammation(13). Oesophageal stenoses occurred in 17% patients with less than two years delayed diagnosis, but in circa 70% with a delay of 15–20 years(17). In young children, the anticipated treatment period is very long – even up to several dozen years. The TWTa analysis can disclose oesophageal wall fibrosis during the routine check-ups. This is an additional benefit of our proposal.
There are some specific features that can be observed during endoscopy in EoE patients: mucosal fragility or oedema, white exudate areas, longitudinal groove and circular folds (transient or persistent; so called oesophageal trachealisation)(7). The whole oesophagus appears as a narrow tube. However, these changes can be seen from the inner aspect and, therefore, there is a need for an additional method to support the endoscopic examination. If untreated, the disease leads to oesophageal fibrosis and fibrostenotic strictures(2,7,18).
To establish the diagnosis, tissue specimens are taken by endoscopic biopsy (2–4 specimens should be assessed during gastroscopy). Eosinophilic infiltration in the mucosa (at least 15 eosinophils in the microscopic field; magnification 400x in light microscopy) is obligatory for diagnosis. Eosinophils in the oesophageal mucosa are present in more than one condition, for example in gastroesophageal reflux disease. Moreover, the same eosinophilic infiltration can be observed in coeliac or Crohn diseases or intestinal inflammations (but usually less than 15 cells). The treatment in those conditions is different, so the differentiation is crucial. In our study, we introduced a simple method to support the supervision, namely changes in TWT.
We chose the weight-dependent reference value, because the normal diameter (an important factor in our new method) of the oesophagus is corelated with the child’s weight, but not with age(13). We decided to base our measurements on TWT, and not on the mucosal, submucosal or muscular layer thickness, due to its simplicity and reliability. For example, during steroid therapy in EoE patients, changes are mainly observed in the increasing mucosal layer thickness, while the other layers are resistant to therapy(19). We decided to choose TWT as a key parameter as more reliable than others. We decided to measure the thoracic part of the oesophagus at Th3-Th7. The usefulness of the mucular layer thickness instead of TWT should be considered and analysed in further studies.
EUS wall measurements may present challenges due to its specific anatomical morphology. When empty, the oesophagus looks like typical tubular viscera (Fig. 1). The folds could be a potential pitfall(20). To exclude these problems, we postulate performing the examination in a standardised oesophageal lumen established previously (13), for example by using the balloon enclosed the US transducer proposed by Rabinowitz(20).
Lastly, we postulate to rely on the square oesophageal wall according to the formula for measuring the square area of the ring/circular belt. The formula for the square area is a product of multiplication n and subtraction of square radiuses of oesophagus lumen and TWT (calculated together) and lumen radius; A = π (Z2–r2). The square area measurements are more reliable, because the risk of mistake is smaller, with four measurements and extraction of the mean value. We hope that this method for measuring the square area of the wall as a ring could be developed and introduced in other radiological examinations, including X-ray, CT and/or MRI, in which the oesophageal wall thickness could be measured.
We agree with Min
AI is still in its early stage, but it is being rapidly put into use. Machine learning is an AI method in which data analysis is focused on the recognition of different patterns. So, in the simplest terms, the AI uses input data to predict the real visceral status, for example oesophageal wall thickness. Deep learning methods provide computer-aided detection of mucosal ulcers on capsule endoscopy images(24). Convolutional neural networks could be used in the analysis of the oesophageal wall status on post-examination images(25). We proposed a set of data to initiate this analysis. An AI algorithm may be created with the proposed data set and further applied for different oesophagus examination methods, among them X-ray, Computer Tomography, Magnetic Resonance Imaging, and US examination. We think that the oesophageal wall thickness could be analysed during routine thoracic cavity examination(26).
Moreover, the newly introduced TWTa method could be introduced for monitoring children after oesophageal burns, surgical reconstructions, connective tissue diseases, and scleroderma. We see a potential benefit of TWTa measurements in the treatment of patients with oesophageal wall injuries after foreign body swallowing and in oesophageal reflux disease resistant to treatment (for example in children with congenital or acquired central nervous system disorders). TWTa could also support the diagnosis of oesophageal motor activity disorders. Lastly, it can be used to support the diagnosis of bone-marrow transplant patients with dysphagia in GRAFT evaluation.
We introduced a simple set of minimal and maximal values of the oesophageal wall thickness square area in children of different weight.
We find TWTa measurements useful in monitoring patients with diagnosed EoE to isolate those with fibrosis risk.
It is difficult to conduct studies in children due to the lack of publications. Also, it is challenging to use the same methods due to variety of methods and perspectives presented in paediatric populations. There are no reliable measurements of applied parameters in children. The oesophageal lumen radius landmark values must be regarded with a tolerable variation of circa 2 mm rather than absolutes (natural fluctuation at a given weight).