A compressive sensing approach for inferring cognitive representations with reverse correlation

Uncovering cognitive representations is an elusive goal that is increasingly pursued using the reverse correlation method, wherein human subjects make judgments about ambiguous stimuli. Employing reverse correlation often entails collecting thousands of stimulus-response pairs, which severely limits the breadth of studies that are feasible using the method. Current techniques to improve efficiency bias the outcome. Here we show that this methodological barrier can be diminished using compressive sensing, an advanced signal processing technique designed to improve sampling efficiency. Simulations are performed to demonstrate that compressive sensing can improve the accuracy of reconstructed cognitive representations and dramatically reduce the required number of stimulus-response pairs. Additionally, compressive sensing is used on human subject data from a previous reverse correlation study, demonstrating a dramatic improvement in reconstruction quality. This work concludes by outlining the potential of compressive sensing to improve representation reconstruction throughout the fields of psychology, neuroscience, and beyond.

Stimulus whitening improves the efficiency of reverse correlation

Human perception depends upon internal representations of the environment that help to organize the raw information available from the senses by acting as reference patterns. Internal representations are widely characterized using reverse correlation, a method capable of producing unconstrained estimates of the representation itself, all on the basis of simple responses to random stimuli. Despite its advantages, reverse correlation is often infeasible to apply because of its inefficiency-a very large number of stimulus-response trials are required in order to obtain an accurate estimate. Here, we show that an important source of this inefficiency is small, yet nontrivial, correlations that occur by chance between randomly generated stimuli. We demonstrate in simulation that whitening stimuli to remove such correlations before eliciting responses provides greater than 85% improvement in efficiency for a given estimation quality, as well as a two- to fivefold increase in quality for a given sample size. Moreover, unlike conventional approaches, whitening improves the efficiency of reverse correlation without introducing bias into the estimate, or requiring prior knowledge of the target internal representation. Improving the efficiency of reverse correlation with whitening may enable a broader scope of investigations into the individual variability and potential universality of perceptual mechanisms.

Reverse Correlation Uncovers More Complete Tinnitus Spectra

Goal: This study validates an approach to characterizing the sounds experienced by tinnitus patients via reverse correlation, with potential for characterizing a wider range of sounds than currently possible. Methods: Ten normal-hearing subjects assessed the subjective similarity of random auditory stimuli and target tinnitus-like sounds ("buzzing" and "roaring"). Reconstructions of the targets were obtained by regressing subject responses on the stimuli, and were compared for accuracy to the frequency spectra of the targets using Pearson's [Formula: see text]. Results: Reconstruction accuracy was significantly higher than chance across subjects: buzzing: [Formula: see text] (mean [Formula: see text] s.d.), [Formula: see text], [Formula: see text]; roaring: [Formula: see text], [Formula: see text], [Formula: see text]; combined: [Formula: see text], [Formula: see text], [Formula: see text]. Conclusion: Reverse correlation can accurately reconstruct non-tonal tinnitus-like sounds in normal-hearing subjects, indicating its potential for characterizing the sounds experienced by patients with non-tonal tinnitus.

Diagnostic Accuracy of Endobronchial Optical Coherence Tomography for the Microscopic Diagnosis of Usual Interstitial Pneumonia

Rationale: Early, accurate diagnosis of interstitial lung disease (ILD) informs prognosis and therapy, especially in idiopathic pulmonary fibrosis (IPF). Current diagnostic methods are imperfect. High-resolution computed tomography has limited resolution, and surgical lung biopsy (SLB) carries risks of morbidity and mortality. Endobronchial optical coherence tomography (EB-OCT) is a low-risk, bronchoscope-compatible modality that images large lung volumes in vivo with microscopic resolution, including subpleural lung, and has the potential to improve the diagnostic accuracy of bronchoscopy for ILD diagnosis. Objectives: We performed a prospective diagnostic accuracy study of EB-OCT in patients with ILD with a low-confidence diagnosis undergoing SLB. The primary endpoints were EB-OCT sensitivity/specificity for diagnosis of the histopathologic pattern of usual interstitial pneumonia (UIP) and clinical IPF. The secondary endpoint was agreement between EB-OCT and SLB for diagnosis of the ILD fibrosis pattern. Methods: EB-OCT was performed immediately before SLB. The resulting EB-OCT images and histopathology were interpreted by blinded, independent pathologists. Clinical diagnosis was obtained from the treating pulmonologists after SLB, blinded to EB-OCT. Measurements and Main Results: We enrolled 31 patients, and 4 were excluded because of inconclusive histopathology or lack of EB-OCT data. Twenty-seven patients were included in the analysis (16 men, average age: 65.0 yr): 12 were diagnosed with UIP and 15 with non-UIP ILD. Average FVC and DlCO were 75.3% (SD, 18.5) and 53.5% (SD, 16.4), respectively. Sensitivity and specificity of EB-OCT was 100% (95% confidence interval, 75.8-100.0%) and 100% (79.6-100%), respectively, for both histopathologic UIP and clinical diagnosis of IPF. There was high agreement between EB-OCT and histopathology for diagnosis of ILD fibrosis pattern (weighted κ: 0.87 [0.72-1.0]). Conclusions: EB-OCT is a safe, accurate method for microscopic ILD diagnosis, as a complement to high-resolution computed tomography and an alternative to SLB.

Rapid non-destructive volumetric tumor yield assessment in fresh lung core needle biopsies using polarization sensitive optical coherence tomography

Adequate tumor yield in core-needle biopsy (CNB) specimens is essential in lung cancer for accurate histological diagnosis, molecular testing for therapeutic decision-making, and tumor biobanking for research. Insufficient tumor sampling in CNB is common, primarily due to inadvertent sampling of tumor-associated fibrosis or atelectatic lung, leading to repeat procedures and delayed diagnosis. Currently, there is no method for rapid, non-destructive intraprocedural assessment of CNBs. Polarization-sensitive optical coherence tomography (PS-OCT) is a high-resolution, volumetric imaging technique that has the potential to meet this clinical need. PS-OCT detects endogenous tissue properties, including birefringence from collagen, and degree of polarization uniformity (DOPU) indicative of tissue depolarization. Here, PS-OCT birefringence and DOPU measurements were used to quantify the amount of tumor, fibrosis, and normal lung parenchyma in 42 fresh, intact lung CNB specimens. PS-OCT results were compared to and validated against matched histology in a blinded assessment. Linear regression analysis showed strong correlations between PS-OCT and matched histology for quantification of tumors, fibrosis, and normal lung parenchyma in CNBs. PS-OCT distinguished CNBs with low tumor content from those with higher tumor content with high sensitivity and specificity. This study demonstrates the potential of PS-OCT as a method for rapid, non-destructive, label-free intra-procedural tumor yield assessment.

Practical application and validation of the 2018 ATS/ERS/JRS/ALAT and Fleischner Society guidelines for the diagnosis of idiopathic pulmonary fibrosis

Background

Accurate diagnosis of idiopathic pulmonary fibrosis (IPF) is essential to inform prognosis and treatment. In 2018, the ATS/ERS/JRS/ALAT and Fleischner Society released new diagnostic guidelines for usual interstitial pneumonitis (UIP)/IPF, adding Probable UIP as a CT category based on prior studies demonstrating this category had relatively high positive predictive value (PPV) for histopathologic UIP/Probable UIP. This study applies the 2018 ATS/ERS/JRS/ALAT and Fleischner Society guidelines to determine test characteristics of CT categories in academic clinical practice.

Methods

CT and histopathology were evaluated by three thoracic radiologists and two thoracic pathologists. Comparison of consensus categorization by the 2018 ATS and Fleischner Society guidelines by CT and histopathology was performed.

Results

Of patients with CT UIP, 87% (PPV, 95% CI: 60–98%) had histopathologic UIP with 97% (CI: 90–100%) specificity. Of patients with CT Probable UIP, 38% (PPV, CI: 14–68%) had histopathologic UIP and 46% (PPV, CI: 19–75%) had either histopathologic UIP or Probable UIP, with 88% (CI: 77–95%) specificity. Patients with CT Indeterminate and Alternative Diagnosis had histopathologic UIP in 27% (PPV, CI: 6–61%) and 21% (PPV, CI: 11–33%) of cases with specificities of 90% (CI: 80–96%) and 25% (CI: 16–37%). Interobserver variability (kappa) between radiologists ranged 0.32–0.81.

Conclusions

CT UIP and Probable UIP have high specificity for histopathologic UIP, and CT UIP has high PPV for histopathologic UIP. PPV of CT Probable UIP was 46% for combined histopathologic UIP/Probable UIP. Our results indicate that additional studies are needed to further assess and refine the guideline criteria to improve classification performance.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12931-021-01670-7.

Distinguishing Tumor from Associated Fibrosis to Increase Diagnostic Biopsy Yield with Polarization-Sensitive Optical Coherence Tomography

PURPOSE

With recent advancements in personalized medicine, biopsies must contain sufficient tumor for histologic diagnosis and molecular testing. However, inadvertent biopsy of tumor-associated fibrosis compromises tumor yield, resulting in delayed diagnoses and/or repeat procedures when additional tumor is needed. The ability to differentiate tumor from fibrosis intraprocedurally during biopsy could significantly increase tumor yield. Polarization-sensitive optical coherence tomography (PS-OCT) is an imaging modality that is endoscope- and/or needle-compatible, and provides large volumetric views of tissue microstructure with high resolution (∼10 μm) while simultaneously measuring birefringence of organized tissues such as collagen. We aim to determine whether PS-OCT can accurately detect and distinguish tumor-associated fibrosis from tumor.

EXPERIMENTAL DESIGN

PS-OCT was obtained ex vivo in 64 lung nodule samples. PS-OCT birefringence was measured and correlated to collagen content in precisely matched histology, quantified on picrosirius red (PSR) staining.

RESULTS

There was a strong positive correlation between PS-OCT measurement of birefringent fibrosis and total collagen content by PSR (r = 0.793; P < 0.001). In addition, PS-OCT was able to accurately classify tumor regions with >20% fibrosis from those with low fibrosis (≤20%) that would likely yield higher tumor content (P < 0.0001).

CONCLUSIONS

PS-OCT enables accurate fibrosis detection and can distinguish tumor regions with low fibrosis. PS-OCT has significant potential for clinical impact, as the ability to differentiate tumor from fibrosis could be used to guide intraprocedural tissue sampling in vivo, or for rapid biopsy adequacy assessment ex vivo, to increase diagnostic tumor yield essential for patient care and research.