evidence-based medicine fellow 趙顗絢

Post on 16-Oct-2021

9 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

Evidence-Based Medicine

Fellow 趙顗絢

Introduction

• OKU

– Ch8 musculoskeletal imaging

T1-weighted

T2-weighted

proton density-weighted (intermediate)

short-tau inversion recovery (STIR)

gradient-echo

– Ch42 spinal tumors

– Ch49 thoracolumbar trauma

Benign or Malignant

• Rothman

– Malignant: ill-defined, marked heterogenous

enhancement, pedicle, paravertebral mass

– Benign: fat present , focal edema, not pedicle,

not mass

Benign or Malignant

• MRI— the modality of choice

– Bone marrow edema (BME)

– Diffusion-weighted imaging (DWI)

– Opposed-phase (chemical shift) imaging

– A single or subset of potential discrimination

Materials & Methods

• Guidelines for conducting diagnostic

research synthesis, systematic reviews,

and meta-analyses

• MOOSE

– Meta-analyses of observational studies

• QUOROM

– Quality of reporting of meta-analyses

Search Strategy

– Through February 2011

– No limits or language restrictions

• MEDLINE

• EMBASE

• SCOPUS

• Cochrane

– Primary and review articles

Study Selection

• Two independent reviewers

• Title and abstracts

• Could not be excluded full text

• No disagreement between the reviewers

Study Inclusion & Exclusion Criteria

• Inclusion criteria

– Original reports: diagnostic performance

– A reference standard for etiology

– Raw data

• Exclusion criteria

– Duplicate articles

– Heterogeneous study population

– Multiple diagnostic modalities

– Technical notes, cases reports

Data Extraction

• Which features were reported

• The frequency

• To group similar features together

• Exclusion: not find enough information

• Exclusion: not emphasize the DDx

MRI features

Quality Assessment

• QUADAS tool

– Quality assessment of diagnostic accuracy

studies

Data Synthesis & Statistical Analyses

• Sensitivity and specificity

• Diagnostic odds ratio

– Pooled with a random-effects model

• if 95% confidence interval did not cross 1

• If the P value was less than 0.05

– Significantly different

Results

Results

• Malignant VCF

– primarily represent: Mets

– <4%: Myeloma

• 9 studies: vertebral level of involvement

• 4 studies: benign & malignant

– 95% thoracolumbar; 5% cervical (malignant)

Quality assessment

• QUADUS

MRI features

• Signal Intensity Characteristics

– Bone Marrow edema

– Complete replacement

– Band-like pattern

– Fluid sign

– DWI

– Contrast

MRI features

• Signal Intensity Characteristics

MRI features

• Morphological Characteristics

– A paraspinal mass

– Epidural mass

– Pedicle vs. Posterior

– Convexity (anterior)

– Convexity (posterior): diffuse vs. focal

– Disc

– Endplate

– All columns

MRI features

• Morphological Characteristics

MRI features

• Quantitative Characteristics

– Conventional pulse sequences (T1, T2, STIR)

contrast-to-noise ratio

signal intensity ratio (SIR)

– Opposed-phase (chemical shift) imaging

– Diffusion-weighted imaging (DWI)

MRI features

• Quantitative Characteristics

– Opposed-phase (chemical shift) imaging: SIR

<0.8: sensitivity 95% specificity 89%<0.65: sensitivity 95% specificity 100%

positive predictive value 100%negative predictive value 95.2%

1.0- and 1.5-T magnets

MRI features

• Quantitative Characteristics

– DWI: apparent diffusion coefficient (ADC)

Fat-suppressed echo planar imaging (EPI)

& higher b values (≧500): ADC<1.5×10-3

MRI features

• Quantitative Characteristics

MRI features

• Other Lesions (Outside of Index VCF)

– Multiple levels of VCF with BME: benign

– A coexisting healed benign VCF: benign

– Other non-characteristic lesions: malignant

MRI features

• Other Lesions (Outside of Index VCF)

Discussion

• Qualitative MRI vs. Quantitative MRI

• Opposed-phase (chemical shift)

– High discrimination capability

• Diffusion: ACS

– Biophysical & technical factors

– Water-specific models

– Absolute diffusion imaging, with EPI

– ADC: 1~1.5mm2/S

• Magnetic resonance spectroscopy

Research synthesis

• Individual studies

• Observer performance assessment

• A combination of various features

• Applying a constellation of criteria

the nature of a compression fracture

– a higher degree of certainty

– reduce interobserver variability

Limitations

• Convenience samples

– Spectrum bias (selection bias)

– Equivocal cases

– More common

• The unit of analysis was per VCF

• Features

• Myeloma and compression fracture

Future studies

• A robust VCF cohort

• Feature refinement

• Standardized MRI techniques

• Multiple readers

• Prediction model development

Thank You

top related