Juhani Pietilä

The DIARETDB1 diabetic retinopathy database and evaluation protocol

Automatic diagnosis of diabetic retinopathy from digital fundus images has been an active research topic in the medical image processing community. The research interest is justified by the excellent potential for new products in the medical industry and significant reductions in health care costs. However, the maturity of proposed algorithms cannot be judged due to the lack of commonly accepted and representative image database with a verified ground truth and strict evaluation protocol. In this study, an evaluation methodology is proposed and an image database with ground truth is described. The database is publicly available for benchmarking diagnosis algorithms. With the proposed database and protocol, it is possible to compare different algorithms, and correspondingly, analyse their maturity for technology transfer from the research laboratories to the medical practice.

Eight-year follow-up of photorefractive keratectomy for myopia.

PURPOSE: We evaluated 8-year results of excimer laser photorefractive keratectomy (PRK) for myopia in terms of stability and late complications. METHODS: Ninety-two myopic eyes of 55 patients were treated with a single-step method using an Aesculap-Meditec MEL 60 excimer laser with a 5.0-mm ablation zone. Treated eyes were divided into three groups according to preoperative refraction: low myopes (≤-6.00 D), medium myopes (-6.10 to -10.00 D), and high myopes (>-10.00 D). RESULTS: Change in myopic regression stabilized in all myopia groups within 12 months, although a small myopic shift occurred up to 8 years after PRK. Mean change in refraction between 2 and 8 years was -0.42 ± 0.48 D for low myopes, -0.37 ± 0.34 D for medium myopes, and -0.41 ± 0.50 D for high myopes. The percentage of eyes within ±1.00 D of emmetropia 8 years after PRK was 78.3% in the low myopia group, 68.8% in the medium myopia group, and 57.1% in the high myopia group. One eye lost 2 lines of best spectacle-corrected visual acuity due to irregular astigmatism. In 13.0% of eyes, a residual trace corneal haze was observed, which had no effect on visual acuity. Apart from the loss of 2 lines of BSCVA in one eye, there were no other late complications during the study period. CONCLUSIONS: The mean change in refraction between 2 and 8 years was less than -0.50 D, regardless of preoperative refraction, and may be attributed to natural age-related refractive change. The appearance of residual corneal haze after 8 years correlated with the amount of myopic correction. PRK was a safe and stable surgical procedure in this group of patients.

Excimer Laser Photorefractive Keratectomy for Hyperopia

OBJECTIVE Excimer laser photorefractive keratectomy (PRK) has been shown to be an effective method in the treatment of refractive errors, especially myopia. We evaluated prospectively the efficacy, predictability, stability, and safety of excimer laser PRK in the treatment of hyperopia. METHODS Thirty-four hyperopic eyes were treated with an Aesculap-Meditec (MEL 60) excimer laser. The patients were divided into two groups. In the low-moderate hyperopia group, baseline spherical equivalent refraction was between +1.50 and +6.00 diopters (D) (mean, +4.20 +/- 1.30 D) and in the high hyperopia group between +6.25 and +9.75 D (mean, +7.70 +/- 1.30 D). Follow-up visits occurred 1, 3, 6, and 12 months after surgery. RESULTS One-year results were available for a total 27 eyes (79%): 15 eyes with low to moderate hyperopia and 12 eyes with high hyperopia. One year after PRK in the low-moderate group, six eyes (40%) had a refractive error within +/- 1.00 D of emmetropia, but in the high hyperopia group only two eyes (17%) were within +/- 1.00 D of emmetropia; three eyes (20%) and one eye (8%) were within +/- 0.50 D, respectively. The stability of the refractive change was better in the low to moderate hyperopia group; in the high hyperopia group there was still some regression after 6 months. At 12 months, 10 eyes (67%) in the low-moderate and one eye (8%) in the high hyperopia group had postoperative uncorrected visual acuity of 20/40 or better. One eye in the low-moderate hyperopia group saw 20/20 without correction. Only one eye lost two lines of spectacle-corrected visual acuity. Haze was more intense in the high hyperopia group, but it did not reduce visual acuity. No vision-threatening complications were observed. CONCLUSIONS When low to moderate hyperopia up to +6.00 D is treated, excimer laser PRK with the Aesculap Meditec MEL60 laser is safe and moderately effective, and refraction stabilizes after 3 months in most eyes. However, PRK is not sufficient to treat high hyperopia in an effective and predictable way.

LASIK flap creation with the Ziemer femtosecond laser in 787 consecutive eyes.

PURPOSE To present the flap characteristics and short-term efficacy and safety of 787 consecutive LASIK procedures with the FEMTO LDV femtosecond laser (Ziemer Ophthalmic Systems) for the treatment of refractive errors. METHODS Seven hundred eighty-seven consecutive eyes of 405 previously non-operated patients were treated with the FEMTO LDV. Intended flap thickness was 110 microm and intended flap diameter varied from 8.5 to 9.5 mm. Refractive treatment was performed with the WaveLight ALLEGRETTO WAVE Concerto 500 Hz excimer laser. All eyes were wavefront-optimized. RESULTS The mean flap thickness, measured by ultrasound pachymetry, was 90.0+/-5.5 microm (range: 67 to 107 microm) in right eyes and 90.1+/-4.6 microm (range: 77 to 106 microm) in left eyes. Mean flap diameter was 9.1+/-0.2 mm (range: 8.4 to 9.9 mm) in right eyes and 9.1+/-0.2 mm (range: 8.0 to 10.0 mm) in left eyes. Increasing flap thickness was correlated with increasing corneal thickness in right eyes and flatter keratometric value K(1) in left eyes. The most common complication was minor bleeding during the procedure (12.7%). All other complications were rare (8.4%), and none prevented further laser ablation. CONCLUSIONS The Ziemer FEMTO LDV laser created thinner LASIK flaps than intended but with a low standard deviation and minimal intraoperative complications.

Corneal Flap Measurements in Laser in situ Keratomileusis Using the Moria M2 Automated Microkeratome

PURPOSE To evaluate accuracy and predictability and factors that influence the dimensions of the laser in situ keratomileusis (LASIK) corneal flap created with the Moria M2 automated microkeratome (Moria SA, Antony, France). METHODS The flap thickness of 454 eyes of 243 consecutive patients was measured using subtraction ultrasonic pachymetry during LASIK with the Moria M2 microkeratome head 130 designed to create a 160-microm-thick flap. Flap dimensions were evaluated and measurements were correlated with preoperative parameters. A stepwise regression analysis was used to determine the factors that influenced actual flap thickness. RESULTS The preoperative spherical equivalent refraction of the 454 eyes ranged from -12.125 diopters (D) to +6.25 D. Patient age ranged from 18 to 57 years (mean age: 31.3 +/- 8.8 years). Mean preoperative keratometric power K1 was 44.31 +/- 1.59 D and K2 was 43.32 +/- 1.54 D. Mean preoperative central comeal thickness was 552.4 +/- 32.5 microm (range: 466 to 665 microm). With an attempted thickness of 160 microm, the Moria M2 flap thickness ranged from 77 to 209 microm (mean: 153.3 +/- 19.0 microm). Mean horizontal flap diameter was 9.2 +/- 0.2 mm and mean hinge length 4.6 +/- 0.3 mm. Increasing flap thickness was found to correlate with increasing preoperative comeal thickness, younger patient age, and flatter preoperative keratometric power K1. CONCLUSIONS Although the standard deviation of the flap thickness was relatively small, remarkable individual variation was noted. Therefore, the intraoperative calculation of the remaining stromal bed is recommended. Furthermore, the consideration of central corneal thickness, patient age, and preoperative keratometry are helpful parameters to avoid too deep ablation.

Bilateral comparison of corneal flap dimensions with the Moria M2 reusable head and single use head microkeratomes.

PURPOSE To compare the Moria (Antony, France) M2 automated microkeratome with the head 130 to a new disposable single use head to evaluate complications, accuracy, and safety of the procedure. METHODS Ninety-eight eyes of 49 consecutive patients were operated with the Moria M2 microkeratome. One eye was operated with the metallic head 130 and the other with a plastic single use head, both designed to create a 160-microm flap. Intraoperative flap dimensions were correlated to preoperative parameters and evaluated 1 month postoperatively. RESULTS With the head 130, mean thickness was 153.3 microm (standard deviation [SD] 13.3, range: 102 to 179 microm). When using a single use head, mean thickness was 148.0 microm (SD 9.8, range: 120 to 170 microm). Occasional iron particles were observed in one eye with both head types. No true epithelial ingrowth was detected in any of the eyes, but epithelial dots at the wound edge occurred in one eye, when using the head 130, but not in the eyes operated with a single use head. CONCLUSIONS On average, both head types created thinner flaps than attempted. Single use heads produced thinner flaps than the head 130. Accuracy in flap thickness in terms of standard deviation was significantly better in single use heads than in the head 130. Single use heads also had fewer microkeratome-related complications. In clinical practice, the single use head was easier to use because no assembly was required. Plastic single use heads also worked more smoothly than the metallic head 130.

Photorefractive keratectomy for -1.25 to -25.00 diopters of myopia

BACKGROUND We evaluated prospectively the efficacy, predictability, stability, and safety of photorefractive keratectomy (PRK) for myopia. METHODS Three hundred sixty-nine eyes of 257 patients were treated with an Aesculap-Meditec MEL 60 excimer laser. Treated eyes were divided into 3 groups: low myopes (-1.25 to -6.00 D), 226 eyes; medium myopes (-6.10 to -10.00 D), 104 eyes; high myopes (-10.10 to -25.00 D), 39 eyes. Follow-up at 12 months was available for 348 eyes (94%). RESULTS One year after surgery the number of eyes within +/- 1.00 D of emmetropia was 182 (86.7%) for low myopes, 43 (40.5%) for medium myopes, and 12 (30.8%) for high myopes. Values for +/- 0.50 D were low: 142 (67.6%), medium: 29 (29.3%), and high: 9 (23.1%). Three eyes with low myopia (1.4%) and 5 eyes with medium myopia (5.1%) lost 2 or more lines of spectacle-corrected visual acuity. None of the high myopes lost 2 or more lines. Uncorrected visual acuity of 20/20 or better was achieved in 82 eyes (39%) with low myopia; 20/40 or better was achieved in 183 eyes (87.1%). Five eyes (5.1%) of medium myopes achieved 20/20 or better; 52 eyes (52.5%) with medium myopia achieved 20/40 or better. Zero eyes with high myopia achieved 20/20 or better; 11 eyes (28.8%) achieved 20/40 or better. CONCLUSION Photorefractive keratectomy proved to be an effective method to correct myopia up to -6.00 D. For myopia greater than -6.00 D, good results were achieved in most eyes when myopia was less than -10.00 D, but efficacy and predictability decrease. To avoid systematic undercorrection, slight overcorrection must be attempted with the Aesculap-Meditec MEL 60 excimer laser for the treatment of myopia.

Fusion of Multiple Expert Annotations and Overall Score Selection for Medical Image Diagnosis

Two problems especially important for supervised learning and classification in medical image processing are addressed in this study: i) how to fuse medical annotations collected from several medical experts and ii) how to form an image-wise overall score for accurate and reliable automatic diagnosis. Both of the problems are addressed by applying the same receiver operating characteristic (ROC) framework which is made to correspond to the medical practise. The first problem arises from the typical need to collect the medical ground truth from several experts to understand the underlying phenomenon and to increase robustness. However, it is currently unclear how these expert opinions (annotations) should be combined for classification methods. The second problem is due to the ultimate goal of any automatic diagnosis, a patient-based (image-wise) diagnosis, which consequently must be the ultimate evaluation criterion before transferring any methods into practise. Various image processing methods provide several, e.g., spatially distinct, results, which should be combined into a single image-wise score value. We discuss and investigate these two problems in detail, propose good strategies and report experimental results on a diabetic retinopathy database verifying our findings.

Repeated photorefractive keratectomy for undercorrection and regression.

PURPOSE Regression (often combined with significant haze) and undercorrection are the most frequent complications after photorefractive keratectomy (PRK) for myopia. PRK retreatment has been used to treat both of these complications. METHODS Sixty-three eyes (55 patients) were reoperated (PRK) because of initial regression or undercorrection. The mean interval between the operations was 17.2 months (range 5 to 37 mo). Residual myopia before retreatment was less than -5.00 D in 62 eyes (mean -2.21 +/- 0.99 D) and was -10.50 D in one eye. The same surgical PRK technique (Aesculap-Meditec MEL 60 excimer laser with 5 or 6-mm ablation zone) was used in both operations. RESULTS One-year results were available for 51 eyes. In the group of eyes with low residual myopia (<-5.00 D), mean refraction 1 year after retreatment was -0.42 +/- 0.96 D and 43 eyes (86%) were within +/-1.00 D of emmetropia. Uncorrected visual acuity of 20/40 or better was achieved in 43 eyes (86%) after 1 year. One eye lost two lines of best spectacle-corrected visual acuity after two procedures because of haze. Mean haze was the same before and 12 months after reoperation (grade 0.39 vs. 0.38). Significant haze (>2) and high regression after the first PRK occurred after reoperation. The one eye with -10.50 D residual myopia after the first PRK showed high regression, and 1 year after the reoperation refraction was -9.50 D with grade 2 haze. CONCLUSIONS Repeated PRK was safe and in most cases effective in treating regression and undercorrection in eyes with low residual myopia after initial PRK. High regression and especially haze after the initial PRK often reappeared after reoperation.

Femtosecond lasers for laser in situ keratomileusis: a systematic review and meta-analysis

Purpose The aim of this study was to review and meta-analyze whether there are differences between reported femtosecond (FS) lasers for laser-assisted in situ keratomileusis (LASIK) in terms of efficacy, predictability, and safety as primary outcomes and corneal flap thickness measurements and pre- and postoperative complications as secondary outcomes. Methods A comprehensive literature search of PubMed, Science Direct, Scopus, and Cochrane CENTRAL Trials Library databases was conducted to identify the relevant prospective randomized controlled trials of FS lasers for LASIK. Thirty-one articles describing a total of 5,404 eyes were included. Results Based on efficacy, IntraLase FS 10 and 30 kHz gave the best results. Based on predictability and safety, there were no differences between various FS lasers. FEMTO LDV and IntraLase FS 60 kHz produced the most accurate flap thicknesses. IntraLase and Wavelight SF200 had the fewest intraoperative complications. IntraLase, Visumax, and Wavelight FS200 had the most seldom postoperative complications. Conclusion There were dissimilarities between different FS lasers based on efficacy and intraoperative and postoperative complications. All FS lasers were predictable and safe for making corneal flaps in LASIK.