Development of lumbar spine MRI referrals vetting models using machine learning and deep learning algorithms: Comparison models vs healthcare professionals

INTRODUCTION

Referrals vetting is a necessary daily task to ensure the appropriateness of radiology referrals. Vetting requires extensive clinical knowledge and may challenge those responsible. This study aims to develop AI models to automate the vetting process and to compare their performance with healthcare professionals.

METHODS

1020 lumbar spine MRI referrals were collected retrospectively from two Irish hospitals. Three expert MRI radiographers classified the referrals into indicated or not indicated for scanning based on iRefer guidelines. The reference label for each referral was assigned based on the majority voting. The corpus was divided into two datasets, one for the models' development with 920 referrals, and one included 100 referrals used as a held-out for the final comparison of the AI models versus national and international MRI radiographers. Three traditional models were developed: SVM, LR, RF, and two deep neural models, including CNN and Bi-LSTM. For the traditional models, four vectorisation techniques applied: BoW, bigrams, trigrams, and TF-IDF. A textual data augmentation technique was applied to investigate the influence of data augmentation on the models' performances.

RESULTS

RF with BoW achieved the highest AUC reaching 0.99. CNN model outperformed Bi-LSTM with AUC = 0.98. With the augmented dataset, the performance significantly improved with an increase in F1 scores ranging from 1% to 7%. All models outperformed the national and international radiographers when compared on the hold-out dataset.

CONCLUSION

The models assigned the referrals' appropriateness with higher accuracies than the national and international radiographers. Applying data augmentation significantly improved the models' performances.

IMPLICATIONS FOR PRACTICE

The outcomes suggest that the use of AI for checking referrals' eligibility could serve as a supporting tool to improve the referrals' management in radiology departments.

Preparation of iron oxide mesoporous magnetic microparticles as novel multidrug carriers for synergistic anticancer therapy and deep tumor penetration

The preparation of mesoporous iron oxides with controllable physiochemical properties for effective therapeutic drug delivery remains a formidable challenge. Herein, iron oxide mesoporous magnetic microparticles (IO-MMMs) were prepared by a modified reverse hard-templating approach using, for the first time, acid-prepared mesoporous spheres (APMS) as the hard silica template. The obtained mesostructures exhibited remarkably high surface area and large pore volumes (S_BET = 240 m²/g and V_pore = 0.55 cm³/g), controllable average sizes, generally uniform morphologies, and excellent biocompatibilities, allowing them to achieve optimal drug release in cancer cells and tumor tissues. IO-MMM carriers were able to co-load high amounts of hydrophilic chemotherapeutic drugs (Dox or Daun) and/or hydrophobic hormonal anticancer drugs (Tam), and release them sustainably in a pH-dependent manner, utilizing the fluorescence of Daun to real-time trace the intracellular drug distribution, and employing Daun/Tam to treat cancer by combined chemo/hormonal therapy. Cytotoxicity assays against different types of cancerous cells showed that the combinatory Daun/Tam@IO-MMM formulation significantly reduced the viability of metastatic MCF7 and KAIMRC1 breast as well as HCT8 colorectal cancer cells, with the least potency towards non-cancerous normal primary cells (up to 10-fold). Electron, flow, and live confocal microscopy imaging confirmed that the loaded vehicles were successfully and differentially uptaken by the different tested cells, gradually releasing their payloads, and causing apoptotic cell death. Importantly, compared to free drugs, Daun/Tam@IO-MMMs displayed enhanced drug accumulation in patient breast primary tumor tissues, deeply penetrating into the tumor region and killing the tumor cells inside. The designed carriers described here, thus, constitute a novel promising magnetic mesoporous smart system that entraps different kinds of drugs and release them in a controlled manner for combinatorial chemo/hormonal cancer theranostics. This multifactorial platform may open new avenues in cancer therapy as efficient synergistic antitumor system through overcoming limitations of conventional cancer therapy.

Characterization and hepatoprotective evaluation of sesquiterpenes and diterpenes from the aerial parts of Juniperus sabina L

Previously, we reported on the hepatoprotective activity of the total extract of Juniperus sabina L. against CCl₄ induced liver toxicity in experimental animals. Biologically directed phytochemical study was conducted to identify the active compounds. Male Wistar rats and the standard drug silymarin were used in the study. Hepatoprotective activity was evaluated via serum biochemical parameters such as aspartate amino transferase (AST), alanine aminotransferase (ALT), gamma glutamyl transpeptidase (GGT), alkaline phosphatase (ALP) and total bilirubin. Tissue parameters including non-protein sulfhydryl groups (NP-SH), malonaldehyde (MDA) and total protein (TP) were also determined. Histopathological study was conducted utilizing Mayer's hematoxylin stain, Periodic Acid Schiff - Hematoxylin (PAS-H) and Masson trichrome technique on light microscope. Electron microscope images were also generated for the study. The activity of the total extract was trapped to the petroleum ether fraction after liquid-liquid fractionation where 51% reduction in the levels of AST, bilirubin and 44% in the levels of ALT were observed. Chromatographic purification of the petroleum ether fraction resulted in the isolation of nine compounds namely: trans-calamenene (1), cadalene (cadalin) (2), epi-cubenol (3), manool (4), calamenene-10β-ol (5), calamenene-10α-ol (6), 4-epi-abietic acid (7), sandaracopimaric acid (8) and isopimaric acid (9). Compounds 1-3, 5 and 6 are belonging to cadinane sesquiterepenes, while compounds 4, 7-9 were of diterpene skeleton. The major compounds were tested for their hepatoprotective effect. Compounds 3 showed marked improvement in the levels of AST and ALT, compound 4 was effective in improving the levels of AST, ALT, GGT, ALP and bilirubin, while compound 7 showed significant improvement in GGT, ALP and bilirubin levels.