18F-FDG-based radiomics and machine learning: a useful help for aortic prosthetic valve infective endocarditis diagnosis?

Introduction

FDG PET/CT allows for a better sensitivity in the prosthetic valve endocartitis (PVE) diagnostic when integrated to ESC 2015 criteria, but visual image analysis results in a weaker specificity and is subject to an inter-observer variability. We therefore aimed to evaluate the interest of quantitative analysis using radiomics and machine learning of FDG PET/CT scans in the PVE diagnostic.

Material and methods

Between 2015 and 2021, patients referred for a FDG PET/CT in our nuclear medicine department with suspected PVE were retrospectively included. The initial development of the model was focused on aortic prosthetic valve (aPV). The aPV was segmented and 31 radiomics features were extracted using the IBSI compliant PyRadiomics framework. Radiomics features were first tested by shuffling 50 times the signal within the aortic segmentation and non-contributive (i.e. identical results within 2×1,96σ over 50 iterations) were excluded. Correlated features were further removed using the variable inflation factor blinded to outcome and remaining features were standardized. Four machine learning algorithms (Ridge and LASSO logistic regression, support vector classifier and random forest) were evaluated and tuned through the use of a training database of patients with aPV included from 2015 to 2019 (excluding positive patient with a mitral and aortic valve). The procedure was further tested through 100 loops on an additional cohort of patients with only aPV included after 2019. ROC curves were subsequently computed and sensitivity was derived based on a fixed specificity of 0.7. Gold standard consisted in an expert consensus from the Endocarditis team. Primary objective was to assess the diagnostic performances of our combined approach using radiomics features and clinical features related to the PET exam (i.e time between aPV implantation and FDG PET/CT, time between antibiotics initiation and FDG PET/CT, extracardiac positive foci, spleen uptake and bone marrow uptake greater than liver uptake).

Results

108 patients were included, for a total of 65 definite PVE and 43 rejected PVE according to the expert consensus. The four algorithms were trained on a total of 68 patients and further tested on a cohort of 40 patients. The performance metrics are reported in the table. Support vector classifier achieved the best scores with an AUC of 0.79±0.01 (sensitivity 0.74±0.03; specificity 0.7). When adding clinical features, AUC was 0.82±0.02 (sensitivity 0.78±0.02; specificity 0.7).

Conclusion

When analyzed with our machine learning-based algorithms, FDG PET/CT reached acceptable diagnostic performances in terms of sensitivity for a specificity corresponding to the results reported by the ENDOPET study (1). These preliminary results obtained on a small test dataset suggest that an artificial intelligence-based algorithm may then guide the final diagnosis especially in this area of subjective visual assessment of PVE.

Funding Acknowledgement

Type of funding sources: None.

Early vs late surgical treatment of radial instability of the thumb metacarpophalangeal joint. A retrospective cohort study

There is no clear evidence in the literature whether treating thumb radial collateral ligament (RCL) injury in the acute phase improves outcome. The purpose of the present study was to compare the clinical and radiological results of RCL repair in acute and chronic settings. Fourteen patients were included. Clinical range of motion (ROM) of the thumb, Kapandji score and radiological parameters were compared pre- and post-operatively to the contralateral uninjured thumb (control group) to evaluate the results of the surgical technique. Patients were then divided into two groups according to early versus late repair and outcomes were compared between the two groups and the control group. Preoperatively, mean spontaneous angle between first metacarpal (M1) and proximal phalanx (P1) (spontaneous M1P1 angle), ulnar stress M1P1 angle, ROM and Kapandji score differed significantly between injured and uninjured sides. Postoperatively these parameters for the injured side improved, reaching values similar to those on the uninjured side, especially with acute phase treatment; late treatment also tended to provide clinical improvement in ROM, Kapandji score and ulnar stress angle, but with significant improvement only for spontaneous deviation of the thumb. This study showed the late and immediate repair of the RCL of the thumb both gave good results, with slightly better outcome with acute phase repair. LEVEL OF EVIDENCE: Therapeutic, Level III.

Understanding the foot's functional anatomy in physiological and pathological conditions: the calcaneopedal unit concept

BACKGROUND

A thorough review of the available orthopaedic literature shows significant controversies, inconsistencies and sparse data regarding the terminology used to describe foot deformities. This lack of consensus on terminology creates confusion in professional discussions of foot anatomy, pathoanatomy and treatment of deformities. The controversies apply to joint movements as well as static relationships between the bones.

DESCRIPTION

The calcaneopedal unit (CPU) is a specific anatomical and physiological entity, represented by the entire foot excepted the talus. The calcaneus, midfoot and forefoot are solidly bound by three strong ligaments that create a unit that articulates with the talus. The movement of the CPU is complex, as it rotates under the talus, around the axis of Henke that coincides with the talo-calcaneal ligament of Farabeuf.This calcaneopedal unit is deformable. It is compared with a twisted plate, able to adapt to many physiological situations in standing position, in order to acheive a plantigrade position.Moreover, the calcaneopedal unit and the talo-tibiofibular complex are interdependent; rotation of the latter produces morphologic modifications inside the former and vice versa.

PURPOSE

This paper is a review article of this concept and of its physiopathological applications.

Positional disappearance of motor evoked potentials is much more likely to occur in non-idiopathic scoliosis

PURPOSE

This study evaluates intraoperative disappearance of motor waveforms related to patient positioning in neurologically asymptomatic patients with spinal deformity.

METHODS

This is a retrospective review of 190 neurologically asymptomatic patients aged seven to 17 years planned for posterior instrumentation under neuromonitoring. There were 159 patients with adolescent idiopathic scoliosis and 31 patients with secondary scoliosis. Patients underwent surgery with transcranial electric stimulation motor evoked potentials (TES-MEPs). In case of abnormal findings, surgery was temporarily discontinued and necessary measures undertaken. In case of permanent signal disappearance surgery was definitively discontinued.

RESULTS

Six patients showed permanent loss of signal during early stages of surgery. These patients had a mean major curve of 64° Cobb angle and a mean thoracic kyphosis (D2 to D12) of 72°. The 184 remaining patients had a mean major curve of 50° Cobb angle and a thoracic kyphosis of 35°. A retrospective descriptive review of the patients' radiographs shows hyperkyphosis to be the common ground between the six secondary scoliosis cases. Gradual preoperative traction maintained during the surgery applied in two of these patients taken back to surgery six months later was associated with maintenance of TES-MEP signals throughout the surgery.

CONCLUSION

This study shows that positional permanent loss of neuromonitoring signals is more likely to occur in patients with secondary scoliosis and hyperkyphosis shown to have sharper spine deformity and suspected to have a more vulnerable spinal cord. Gradual skeletal traction performed in two of these patients and maintained during surgery showed promising results.

LEVEL OF EVIDENCE

IV.

Selecting the lowest instrumented vertebra in adolescent idiopathic scoliosis: Comparison of the Lenke, Suk, and Dubousset criteria

BACKGROUND

Selection of the lowest instrumented vertebra (LIV) in patients undergoing selective fusion for Lenke type 1 or 2 adolescent idiopathic scoliosis (AIS) varies widely across centres around the world.

HYPOTHESIS

Lenke, Suk, and Dubousset criteria show moderate agreement for LIV selection.

METHODS

Sixty-eight patients with Lenke type 1 or 2 AIS managed by selective posterior fusion and followed-up for at least 2 years were included in a retrospective observational study. Agreement among Lenke, Suk, and Dubousset criteria for LIV selection was assessed. For surgery, the LIV was selected based on Dubousset criteria. Retrospectively, in each patient, the LIV selected by the Lenke and Suk criteria sets was identified on the preoperative images. The patients were then divided into two groups based on whether the Dubousset LIV was identical versus more distal than the LIV identified retrospectively by the Lenke or Suk criteria. The primary evaluation criterion was coronal balance.

RESULTS

The LIVs selected by the Lenke, Suk, and Dubousset criteria were identical in 57% of cases. The LIV selected by the Dubousset criteria were identical to that selected by the Lenke or Suk criteria in 70% of patients. No significant between-group differences were found for any of the evaluation criteria assessed preoperatively, postoperatively, or at last follow-up.

DISCUSSION

Agreement among the Lenke, Suk, and Dubousset criteria was moderate, confirming the working hypothesis. No coronal malalignment developed in the patients whose actual LIV was distal to the LIV selected by the Lenke or Suk criteria, supporting the validity of Dubousset criteria for LIV selection. When selecting the LIV, all three criteria sets should be assessed. The LIV is the vertebra selected by all three if they agree or by the Dubousset criteria if they do not.

LEVEL OF EVIDENCE

IV, retrospective cohort study.

Sclerotherapy using Surgiflo and alcohol: a new alternative for the treatment of aneurysmal bone cysts

PURPOSE

This study aims to review the results of sclerotherapy using Surgiflo in addition to alcohol in primary aneurysmal bone cysts (ABC).

METHODS

A total of 16 consecutive patients with histologically confirmed diagnosis of primary ABC were treated by percutaneous Surgiflo and alcohol injection at our institution. Clinical follow-up included the assessment of pain, swelling, limping and functional impairment. Radiological parameters included tumour volume, physis-cyst distance, thickness of cyst cortex, and presence of intracystic septations. Mean follow-up was 35,6 months (24-71 months). Treatment was considered successful when the cyst volume decreased by a minimum of 10%, the bone cortex became thicker, and the distance to physis increased.

RESULTS

Mean age at presentation was 9.5 years (5.16-13.84 years). All ABC's were primary and all patients underwent a single Surgiflo and alcohol session except for two (12.5%) who required a second session. All patients had a good clinical result at final follow-up. Satisfactory cyst healing was achieved in 11 cases according to radiological parameters. Tumour volume decreased from a mean of 122 cm³ (111 to 133) before injection to 86 cm³ (76 to 96) at last follow-up (p < 0.01). Physis-cyst distance increased from a mean of 1 cm (0.1 to 2) to 2.1 cm (0.5 to 4) at last follow-up (p < 0.01). Cortical thickness improved from 1 mm (0.5 to 1.5) to 2 mm (1 to 3.5) at last follow-up (p < 0.01).There were no treatment related complications. Surgery was performed in one patient having a C3 vertebra ABC after developing quadriparesis due to tumour progression.

CONCLUSION

Sclerotherapy using Surgiflo and alcohol may be used as an efficient, safe and minimally invasive alternative for the treatment of primary ABCs.