Secondary Logo

Journal Logo

Illumination, optics, and clinical performance of a hand-held magnified visual inspection device (AviScope™): a comparison with colposcopy

Sellors, John W. MD*; Winkler, Jennifer L. MPH*; Kreysar, Douglas F. MS

JAIDS Journal of Acquired Immune Deficiency Syndromes: October 2004 - Volume 37 - Issue - p S160–S166
SUPPLEMENT ARTICLE
Free
SDC

Objective Colposcopy is the reference standard for a visual inspection device in terms of illumination, optics, and clinical performance. A hand-held magnification device such as the AviScope, developed as an alternative to naked eye visual inspection with acetic acid, is also of interest as a low-cost, more portable alternative to a colposcope in low-resource settings within the context of cervical cancer prevention programs and for microbicides research.

Design A performance comparison of the AviScope, visual inspection with acetic acid, and three colposcopes.

Methods An analysis was carried out of the optics and illumination of the colposcopes and the AviScope prototype and of the evidence for the reproducibility and clinical accuracy of visual methods for histologically confirmed high-grade cervical intraepithelial neoplasia. Using the findings, the feasibility of increasing the performance and reducing the cost of a hand-held scope was examined.

Results Published studies have found the AviScope to have slightly higher sensitivity than visual inspection with acetic acid (60.7 versus 55.7%, P <0.05) without loss of specificity. Despite the variability among the colposcopes, the field of view, resolution, depth of field, magnification, and quality of the illumination beam pattern and spectral output exceeded that of the AviScope. The clinical sensitivity and specificity of colposcopy were higher than that of the AviScope. The availability of improved materials for optics and illumination suggests that a hand-held scope with enhanced performance is feasible.

Conclusion Although the performance of the AviScope prototype was suboptimal compared with the colposcopes, it appears possible to design a hand-held magnification device at a reasonable price with better optics and illumination.

*Program for Appropriate Technology in Health (PATH), Seattle, WA, USA

Radiant Imaging, Inc., Duvall, WA, USA

Correspondence to John W. Sellors, MD, 1455 NW Leary Way, Seattle, WA 98107, USA. Tel: +1 206 285 3500; fax: +1 206 285 6619; e-mail: jsellors@path.org

Presented at the Assessing Inflammation and Epithelial Integrity in Vaginal Product Research Meeting, Punta Cana, Dominican Republic, November 19-21, 2003.

The colposcope was introduced by Hinselmann in 19251 and is now in common use for magnified stereoscopic examination of the cervix, vagina, and vulva. Modern colposcopes are sometimes referred to as field microscopes because they have a magnification range of approximately 3-27× and a self-contained light source, and they are, to some extent, portable. Because a colposcope is a relatively expensive piece of equipment, requiring maintenance and an operator with advanced training, colposcopy is usually reserved for the diagnostic examination of women at increased risk of abnormality who have been referred because of a positive screening test [cytology, visual inspection with acetic acid (VIA), etc.] or clinical suspicion of other abnormalities.2 Colposcopic diagnosis is based on the presence of epithelial abnormalities caused by trauma, inflammation, infection, and neoplasia that are observed before or after the application of normal saline, dilute acetic acid, or Lugol's iodine. The visual diagnosis of genital tract neoplasia or other abnormalities can be confirmed pathologically by colposcopically directed biopsy. Except for its relatively high cost, the colposcope is considered the ‘ideal’ level of performance for any visual inspection device in terms of illumination and optical characteristics.

Over several years, the Program for Appropriate Technology in Health (PATH) has conducted research and development of a low-cost, hand-held, 4* magnification scope with a green light source, the aided visual inspection device (AviScope; O'Ryan Industries, Vancouver, WA, USA). The AviScope weighs 350 g and measures 9.5 cm long, 9.5 cm wide, and 6.0 cm high. It is monocular, small enough to fit in a pocket, and has light-emitting diodes (LEDs) powered by self-contained, rechargeable batteries. A collaboration was formed with a private sector manufacturer (O'Ryan Industries) to provide prototypes that were used to test the performance of the device under field conditions in less developed and developed countries during the past 5 years.3–5 Initially intended as a primary screening device for the detection of cervical precancerous lesions within the context of a cervical cancer prevention program, the AviScope is also of interest as a replacement for the colposcope and as a device for confirming the presence of cervical lesions in women who are positive by VIA or visual inspection with Lugol's iodine (VILI). The use of a magnification device to assist the visualization of the cervix during VIA is sometimes referred to as visual inspection with acetic acid and magnification (VIAM). Although we limited our investigation to the AviScope, several other devices for VIAM exist, such as the gynoscope, which does not contain its own light source,6 the magnivisualizer, which does have a light source,7 and a simple hand-held lens.8

Based on the same principles used with a colposcope and the AviScope, VIA is performed with the naked eye, observing the cervix after the insertion of a vaginal speculum, before and then 1–2 min after the application of 3-5% dilute acetic acid. Any abnormalities and acetowhite areas are assessed for their importance, and the examination is categorized as VIA-negative, VIA-positive, or suspicious for cancer.9,10 Similarly, VILI is performed by applying Lugol's iodine solution, and the cervix is immediately assessed as VILI-negative, VILI-positive, or suspicious for cancer.10 The acetowhite and reduced iodine uptake reactions are not specific for neoplasia, and may be caused by trauma or inflammation of the epithelium of the anogenital tract.2

To understand the performance gap between the colposcope and simpler visual inspection methods, we compared the clinical performance of VIA, VILI, and the AviScope prototype with the colposcope. The optics and illumination characteristics of the AviScope and three typical colposcopes were also examined, and together with the field experience with the AviScope prototype, the preliminary concept for a hand-held magnified visual inspection device with enhanced performance and at a reasonable cost is given.

Back to Top | Article Outline

METHODS

The assessment of accuracy of the various approaches to visual inspection of the cervix is based on published studies that have methodological rigor by virtue of using the population-based recruitment of large numbers of women in low-resource settings and the avoidance of verification bias by using colposcopy and histology as a reference standard in all women. Reproducibility assessments are based on published studies that use photographic records of cervical examinations as the basis for testing observer agreement.

PATH and a private sector engineering laboratory (Radiant Imaging, Inc., Duvall, WA, USA) conducted a comparison of the optics and illumination of the AviScope prototype and three colposcopes from Germany, India, and the United States, respectively referred to as colposcopes 1, 2, and 3. Optical characteristics include resolution, field of view (FOV), magnification, and depth of field (DOF). Resolution is a measure of the smallest object that can be viewed by an optical system, and in this case is measured as the number of line pairs per millimeter (lp/mm) that can be distinguished with a device used by an observer with 20/20 eyesight. The higher the resolution, the more one can distinguish features of what is being viewed. The resolution is tested against a 1963A National Bureau of Standards resolution pattern containing five pairs of lines and many different lp/mm up to a maximum of 18 lp/mm. The target pattern is placed at the focus of the device, and the best resolution possible is measured at the center of the field. The target is translated horizontally across the field, and the resolution at 50% and 90% of the field is measured. FOV is the width (mm) of area that can be viewed using an optical system, at the object distance when the image is in focus. Magnification is the increase in apparent object size. DOF is the ability of an optical system to image objects at different distances from the optical system, and this is important when using a hand-held optical system or a surgical instrument when viewing an object. A higher DOF means that more of a contoured surface is visible, and an image is less likely to go out of focus if the optical device is moved slightly. DOF is measured by translating the National Bureau of Standards resolution target along the optical axis of the system and measuring distances where different levels of resolution can be achieved.

The beam pattern created by the illumination system of each device is tested using a ProMetric 1401–1 CCD camera photometric measurement system (Radiant Imaging). Each device is set up to shine its illumination pattern on a white, diffuse screen. The distance between the white screen and the device is set to the focal distance of the device. The ProMetric system captures an image of the beam pattern on the screen and converts that image into photometric units. The beam patterns are represented in colors in which red indicates a region of higher light intensity and blue indicates a region of lower light intensity. The cross-section of the beam pattern is shown to indicate the uniformity of illumination across the field. Uniformity is important because this allows one to distinguish between light and dark tissue without the introduction of artificially created effects caused by the source of illumination. A flat horizontal line would indicate a very uniform illumination, whereas peaks and valleys indicate areas of high and low brightness, respectively, (non-uniform illumination). The illumination beam pattern is described in terms of the total amount of light (expressed as lumens), the diameter of the beam (mm), maximum illumination (foot-candle), and uniformity, which is a subjective rating of how evenly the light is spread across the beam. In order to compare the color of the light emitted from each device, the spectral power distribution (power versus wavelength) of each illumination pattern is measured. The beam pattern is measured as reflected from a white, diffuse screen, and the normalized spectral power distribution of light is calculated for each light tested (expressed as arbitrary units). Beam pattern is important because of user preference, and could potentially be important for the perceived contrast of an object. The white, diffuse screen will change slightly the spectral power characteristics of the illumination system because the reflectivity of the screen is not constant compared with wavelength. The error introduced by this method is considered to be slight, so that this method should be adequate for the purposes of comparing the devices.

Based on the results of field performance, optics, illumination, and current materials, the preliminary concept and set of specifications for a hand-held magnification device with suitable characteristics for use in low-resource settings has been developed.

Back to Top | Article Outline

RESULTS

Accuracy of visual methods for the detection of cervical neoplasia

In a population-based, cross-sectional survey of 1997 women aged 35–45 years in Shanxi Province, China,11 all of the participants were colposcopically assessed and biopsied at least four times on the ectocervix, and all had an endocervical curettage in order to reduce verification and other biases that might affect the estimation of the performance of colposcopy and VIA. Sensitivity and specificity for high-grade cervical intraepithelial neoplasia (CIN) or cancer (prevalence 4.3%, 86/1997) were 81% (95% confidence interval [CI] 72%, 89%) and 77% (95% CI 75%, 78%). In the same study, VIA had a sensitivity and specificity of 71% (95% CI 70%, 80%) and 74% (95% CI 72%, 76%). In another cross-sectional survey of 4444 women aged 25–65 years in Kerala, India,12 the prevalence of CIN 2/3 was 3.4% (149/4444); the sensitivities of VIA and VILI were 82.6% and 87.2%, and the specificities, 86.5% and 84.7%, respectively. A cross-sectional survey comparing VIA, VIAM (AviScope), and conventional cytology was conducted in Kolkata, India, with 5881 women aged 30–64 years.5 The sensitivities for CIN 2/3 (prevalence 2.1%, 122/5881) for VIA, AviScope, and cytology (CIN 1 or higher cutoff) were 55.7%, 60.7%, and 29.5%, with specificities of 82.1%, 83.2%, and 92.3%, respectively. The AviScope did have a small but significantly higher sensitivity than VIA (P <0.05) without a loss of specificity.

Back to Top | Article Outline

Reproducibility of visual methods for the detection of cervical neoplasia

Using cervical photographs taken after the application of dilute acetic acid, it has been possible to conduct observer agreement studies of visual methods. Three expert colposcopists showed poor to good intraobserver and interobserver agreement on the border characteristics (range of interobserver kappa 0.13–0.41, and intraobserver kappa 0.26–0.58) and the color of acetowhitening (range of interobserver kappa 0.21–0.47, and intraobserver kappa 0.34–0.75), but had excellent agreement on the site of the lesion and where they would take a biopsy (raw agreement 95.3%, 143/150).13 In contrast to the latter study, in which photographs were enlarged by projecting them onto a screen, in a study using cervical photos that were approximately normal in size,14 three experienced teachers of VIA had good interobserver agreement (kappa 0.57, 95% CI 0.48, 0.66) on whether the images were VIA-negative, VIA-positive, or suspicious for cancer. The suboptimal reproducibility of colposcopy and VIA was consistent with what has been shown for agreement on cervical cytology and histology.15 To our knowledge, there has not been a study of the reproducibility of visual assessment using the AviScope.

Back to Top | Article Outline

Key attributes of cervical neoplastic lesions

Prospective research in a Canadian colposcopy clinic on the predictive validity of visual signs has shown that among the characteristics routinely evaluated for an abnormal transformation zone (borders, degree of acetowhitening, abnormal blood vessels), acetowhitening has the most predictive value and performs as well as all three signs combined.16 It is intuitive that the size of a lesion affects the sensitivity of both colposcopy and VIA, and research in China has confirmed this.11 Colposcopy has a sensitivity of 65% (95% CI 47%, 79%) if the lesion only involves one quadrant of the cervical surface, and 100% if more surface is involved. VIA has a sensitivity of 64% (95% CI 47, 79) if one quadrant is involved, 67% (95% CI 41%, 87%) if two are involved, 100% (95% CI 74%) if three are involved, and 82% (95% CI 57%, 96%) if four quadrants are involved.

Back to Top | Article Outline

Field performance of the AviScope

Experience with the use of the AviScope by various levels of healthcare providers has accumulated over several years in several low-resource field settings in South Africa, Peru, India, Mexico, and Kenya. Providers like the way the device can be held easily in one hand, whereas some have asked about future versions having an optional support that allows hands-free operation if desired. Some feel that the light should be brighter and the magnification higher. The usual length of use of the illumination system after charging the batteries is 3–5 h, which providers find adequate for use in busy, one-day clinics. Charging the self-contained batteries takes 1–3 h and is achieved by simply connecting the AviScope to the small external charger that is plugged into the electrical mains. Although the plastic casing is fairly resilient, breakage in the casing or internal circuitry does occur rarely when the AviScope is dropped. The casing is not waterproof, and care needs to be taken not to immerse the AviScope in fluid such as water or acetic acid. When failures do occur preventing the use of the AviScope, it can either be repaired locally or mailed back to PATH for servicing. It is easy to mail because the dimensions are 9.5 cm long, 9.5 cm wide, and 6 cm high, and the weight is 350 g. Based on the experience of PATH in several low-resource settings, with several types of colposcopes and the AviScope, colposcopes are more difficult and costly to acquire, store, transport, maintain, and repair.

Back to Top | Article Outline

Optics and illumination of the AviScope and colposcopes

The scopes are compared on FOV, magnification, DOF, and focal distance in Table 1. The focal distances are all within a range (260–305 mm) that would allow the use of instruments for procedures while using the device. The AviScope has the lowest resolution, making it difficult to distinguish fine features that would be seen easily by the colposcopes (Table 2). Colposcopes 1 and 2 have the best DOF and the AviScope the lowest (5 mm). Figures 1 and 2 illustrate the contrast between the AviScope and colposcope 2 as to how well the optics of each device maintain focus away from the center of the FOV. The total amount of light from the AviScope is a fraction of that from the colposcopes, and the uniformity is rated as poor for the AviScope and fair for colposcopes 2 and 3 (Table 3). It is also noted that the green filter on colposcopes blocks at least half of the light. Illumination beam patterns of colposcope 3 and the AviScope are shown in Figures 3 and 4. A flat horizontal line would indicate a very uniform illumination, whereas the patterns of both scopes indicate areas of high and low brightness (non-uniform illumination). The normalized spectral power distribution of light from all of the scopes and two flashlights is shown in Figure 5.

FIGURE 1

FIGURE 1

FIGURE 2

FIGURE 2

FIGURE 3

FIGURE 3

FIGURE 4

FIGURE 4

TABLE 1

TABLE 1

TABLE 2

TABLE 2

TABLE 3

TABLE 3

FIGURE 5

FIGURE 5

Back to Top | Article Outline

Feasibility of a new hand-held magnifying scope with illumination

The preliminary concept and set of specifications for a new hand-held scope are derived based on lessons learned and currently available materials. First, the optics of the imaging system could be composed of two cemented achromats, image correcting prisms, and an additional eyepiece lens, held in place within a plastic housing. The use of a new optics system would allow an improvement in resolution, and magnification could be increased up to 8×. The cost of the optics is estimated at US$60, which is a fraction of the current cost of the optics in the AviScope.

Second, to reduce the cost of the illumination optics, incandescent light sources could be considered. Initial calculations show that using an AA flashlight with a green filter could produce approximately 4 lumens. Using two incandescent bulbs, a total of 8 lumens could be achieved, nearly double the lumens achieved by the AviScope (Table 3). Although an initial costing is not complete, the use of something resembling an AA flashlight that retails for US$6–10 is far less costly than the 24 green LEDs and four white LEDs in the AviScope. A further benefit of such a design approach is that a user could switch from white to green illumination by interposing a green filter. In addition to the potential cost saving of switching to incandescent lamps, the possibility exists that an LED illumination solution could be found for significantly less cost than the original LED cost. This is due to the fact that significant advances have been made in the past few years in LED efficiency and light output. Compared with LEDs, incandescent lamps generally have a much shorter life, and the difference in the spectral power distribution of their beam patterns has design implications.

Back to Top | Article Outline

DISCUSSION

The original concept of the AviScope was an alternative to VIA. Based on one methodologically rigorous study of primary screening, the sensitivity of detection of high-grade CIN only slightly increased from 55.7% with VIA to 60.7% with the AviScope.5 In addition, there is no evidence supporting the use of the AviScope as a confirmatory test after primary screening using cytology.4 In the evaluation of the performance of any new test or device, reproducibility should also be assessed, as well as accuracy. Although there is no direct evidence, there is no reason to suspect that the AviScope would be markedly different from the mediocre reproducibility of VIA, colposcopy, and pathology for the detection of cervical neoplasia.13–15 The information from the field studies of the AviScope shows that the physical characteristics, durability, and electrical power system are acceptable, despite the prototypical nature of the device and the use of small-scale, non-routine production methods in its construction. Providers' feedback on the need for a stronger light and higher magnification and resolution agrees with the findings of the optics and illumination study of the AviScope. In low-resource settings, a hand-held device such as the AviScope has many practical advantages over the colposcope that are unrelated to optics or illumination. These advantages include portability, durability, non-reliance on a ready source of mains electricity, ease of repair, shipping, and maintenance.

Although the AviScope was never intended to compete with a colposcope, this alternative use seems to be of great interest at present. In comparison with colposcopes, the AviScope has the weakest light source and the worst optical performance with respect to magnification, DOF, and resolution. The shallow DOF characteristics would make the AviScope the most difficult to keep focused on a target. It is assumed that any device that can be used for magnified examination of the cervix will also allow visualization of the vaginal and vulvar surfaces. Hand-held devices can be mounted on a support for hands-free operation like a colposcope. Images from a hand-held scope could be captured by camera, providing a transfer optic (one or two lenses plus a housing) is attached to image the object to the camera, rather than the human eye.

Improved performance at a reasonable cost appears to be feasible, given the preliminary concept and set of specifications. Further research should be conducted on the optimal performance targets for a new hand-held scope, and pricing information on different illumination systems needs to be gathered.

Back to Top | Article Outline

ACKNOWLEDGEMENTS

This work was supported by the Bill and Melinda Gates Foundation through the Alliance for Cervical Cancer Prevention, the United States Agency for International Development (USAID) under HealthTech Technologies for Health Project, Cooperative Agreement no. HRN-A-00-96-90007, and a National Cancer Institutes SBIR grant, no. NCI.2 R44 CA65313-02A1.

Back to Top | Article Outline

REFERENCES

1. Hinselmann H. Verbesserung der Inspektion-smoglichkeiten von Vulva, Vagina und Portio. Munchner Med Wochenschr, 1925;72:1733–1742.
2. Sellors J, Sankaranarayanan R. Colposcopy and treatment of cervical intraepithelial neoplasia: a beginners' manual. Lyon, France: IARC Press; 2003.
3. Denny L, Kuhn L, Pollack A, et al. Direct visual inspection for cervical cancer screening: an analysis of factors influencing test performance. Cancer, 2002;94:1699–1707.
4. Winkler JL, Tsu VD, Bishop A, et al. Confirmation of cervical neoplasia using a hand-held, lighted magnification device. Int J Gynaecol Obstet,2003;81:35–40.
5. Basu P, Sankaranarayanan R, Mandal R, et al. Visual inspection with acetic acid and cytology in the early detection of cervical neoplasia in Kolkata, India. Int J Gynecol Cancer, 2003;13:1–7.
6. Abrams J. A preliminary study of the gynoscope: an adjunct to cytologic screening of the cervix. Am J Gynecol Health, 1990;4:27–33.
7. Parashari A, Singh V, Sehgal A, et al. Low-cost technology for screening uterine cervical cancer. Bull WHO, 2000;78:964–967.
8. Denny L, Kuhn L, Pollack A, et al. Evaluation of alternative methods of cervical cancer screening for resource-poor settings. Cancer, 2000;89: 826–833.
9. Sellors J, Lewis K, Kidula N, et al. Screening and management of precancerous lesions to prevent cervical cancer in low-resource settings. Asian Pac J Cancer Prev, 2003;4:277–280.
10. Sankaranarayanan R, Wesley R. A practical manual on visual screening for cervical neoplasia. Lyon, France: IARC Press; 2003.
11. Belinson JL, Pretorius RG, Zhang WH, et al. Cervical cancer screening by simple visual inspection after acetic acid. Obstet Gynecol, 2001;98: 441–444.
12. Sankaranarayanan R, Wesley R, Thara S, et al. Test characteristics of visual inspection with 4% acetic acid (VIA) and Lugol's iodine (VILI) in cervical cancer screening in Kerala, India. Int J Cancer, 2003; 106: 404–408.
13. Sellors JW, Nieminen P, Vesterinen E, et al. Observer variability in the scoring of colpophotographs. Obstet Gynecol, 1990;76:1006–1008.
14. Sellors JW, Jeronimo J, Sankaranarayanan R, et al. Assessment of the cervix after acetic acid wash: inter-rater agreement using photographs. Obstet Gynecol, 2002;99:635–640.
15. Stoler MH, Schiffman M. Atypical Squamous Cells of Undetermined Significance - Low-grade Squamous Intraepithelial Lesion Triage Study (ALTS) group. Interobserver reproducibility of cervical cytologic and histologic interpretations: realistic estimates from the ASCUS-LSIL Triage Study. JAMA, 2001;285:1500–1505.
16. Shaw E, Sellors J, Kaczorowski J. Prospective evaluation of colposcopic features in predicting cervical intraepithelial neoplasia: degree of acetowhitening most important. J Lower Genital Tract Dis, 2003;7: 6–10.
Keywords:

Cervical neoplasia; colposcopy; inflammation; microbicides; visual examination

© 2004 Lippincott Williams & Wilkins, Inc.