To retrospectively analyze the technical and long-term clinical outcome of angioplasty and stenting using the Venovo™ venous stent for the treatment of malignant and benign superior vena cava (SVC) occlusive disease. Consecutive patients treated with the Venovo™ venous stent for SVC occlusive disease were included. SVC obstruction symptoms were classified according to the Kishi score. The Wilcoxon signed-rank test was used for testing significance of changes. Technical success, defined as correct placement of the stent, completely covering and re-expanding the obstruction, between groups was tested using the Fisher exact test. Overall survival was calculated using the Kaplan–Meier method. Fifty-five patients underwent stent insertion for symptomatic benign (n = 13; 24

Beware unusual presentations of more common disease entities, as in this interactive case report http://ow.ly/qj7f30eVFsp.

Summary Purpose: To investigate the relationship between the dynamic parameters (Ki) and static image-derived parameters of 68Ga-DOTATOC-PET, to determine which static parameter best reflects underlying somatostatin-receptor-expression (SSR) levels on neuroendocrine tumours (NETs). Patients, methods: 20 patients with metastasized NETs underwent a dynamic and static 68Ga-DOTATOC-PET before PRRT and at 7 and 40 weeks after the first administration of 90Y-DOTATOC (in total 4 cycles were planned); 175 lesions were defined and analyzed on the dynamic as well as static scans. Quantitative analysis was performed using the software PMOD. One to five target lesions per patient were chosen and delineated manually on the baseline dynamic scan and further, on the corresponding static 68Ga-DOTATOC-PET and the dynamic and static 68Ga-DOTATOC-PET at the other time-points; SUVmax and SUVmean of the lesions was assessed on the other six scans. The input function was retrieved from the abdominal aorta on the images. Further on, Ki was calculated using the Patlak-Plot. At last, 5 reference regions for normalization of SUVtumour were delineated on the static scans resulting in 5 ratios (SUVratio). Results: SUVmax and SUVmean of the tumoural lesions on the dynamic 68Ga-DO-TATOC-PET had a very strong correlation with the corresponding parameters in the static scan (R²: 0.94 and 0.95 respectively). SUVmax, SUVmean and Ki of the lesions showed a good linear correlation; the SUVratios correlated poorly with Ki. A significantly better correlation was noticed between Ki and SUVtumour(max and mean) (p < 0.0001). Conclusions: As the dynamic para meter Ki correlates best with the absolute SUVtumour, SUVtumour best reflects underlying SSR-levels in NETs.

PURPOSE:To investigate the relationship between the dynamic parameters (Ki) and static image-derived parameters of 68Ga-DOTATOC-PET, to determine which static parameter best reflects underlying somatostatin-receptor-expression (SSR) levels on neuroendocrine tumours (NETs). PATIENTS, METHODS:20 patients with metastasized NETs underwent a dynamic and static 68Ga-DOTATOC-PET before PRRT and at 7 and 40 weeks after the first administration of 90Y-DOTATOC (in total 4 cycles were planned); 175 lesions were defined and analyzed on the dynamic as well as static scans. Quantitative analysis was performed using the software PMOD. One to five target lesions per patient were chosen and delineated manually on the baseline dynamic scan and further, on the corresponding static 68Ga-DOTATOC-PET and the dynamic and static 68Ga-DOTATOC-PET at the other time-points; SUVmax and SUVmean of the lesions was assessed on the other six scans. The input function was retrieved from the abdominal aorta on the images. Further on, Ki was calculated using the Patlak-Plot. At last, 5 reference regions for normalization of SUVtumour were delineated on the static scans resulting in 5 ratios (SUVratio). RESULTS:SUVmax and SUVmean of the tumoural lesions on the dynamic 68Ga-DOTATOC-PET had a very strong correlation with the corresponding parameters in the static scan (R²: 0.94 and 0.95 respectively). SUVmax, SUVmean and Ki of the lesions showed a good linear correlation; the SUVratios correlated poorly with Ki. A significantly better correlation was noticed between Ki and SUVtumour(max and mean) (p < 0.0001). CONCLUSIONS:As the dynamic parameter Ki correlates best with the absolute SUVtumour, SUVtumour best reflects underlying SSR-levels in NETs.

Background ESMO consensus recommends EGFR mutation testing in never/former light smokers (<15 pack-years) or patients with non-squamous NSCLC. The aim of this work was to determine the frequency and clinical predictors of EGFR mutations, and the role of specimen sampling tests, in Caucasian standard practice setting. Methods We screened 297 patients according to this consensus. Mutational analysis of EGFR was performed using the Therascreen EGFR RGQ PCR mutation kit. Clinical and pathological correlative data were collected. Results An EGFR activating mutation was found in 32 patients (11%), twelve exon 19 deletions, two exon 18 and eighteen exon 21 point mutations. Most were in females, but half were in smokers. Negative TTF-1 staining had a very strong negative predictive value (all except one patient had TTF-1 positive adenocarcinoma). Both biopsies as well as cytology specimens (mainly EBUS-TBNA) did well: 24 mutations in 213 biopsy samples (11.2%) and 8 in 84 cytology samples (9.5%), respectively. The Therascreen acted as a sensitive test in all types of samples: 7 activating mutations were found in samples rated to have <5% of tumour cells, and there were only 4 test failures in the whole series. Conclusion In this Caucasian standard practice NSCLC cohort, tested according to the ESMO consensus, activating EGFR mutation occurred in 11% of the patients. Half of these were in former/current smokers. With our sampling technique and use of the Therascreen kit, EBUS-TBNA cell blocks performed as good as biopsies.

Objective: To evaluate the effectiveness of inpatient brief counselling by a smoking cessation nurse compared to usual care (no advice). Methods: The subjects ( n = 381, 245 men and 136 women) studied were in-patients, in four Flemish University Hospitals, who were daily smokers. Patients were randomised between 2005 and June 2006. Patients were allocated to an experimental group (EG) or to a control group (CG). Allocation and smoking cessation interventions of patients were stage-matched according to their stage of change as defined by Prochaska and Diclemente. Smoking cessation advice was administered by a qualified smoking cessation nurse. Results: The six-month self-reported continuous abstinence in the EG in 28/178 patients (15.7%) compared to the CG where 14/180 patients were abstinent (7.7%) was significantly better. The effect was most pronounced in the subgroup over 40 years old in the preparation and action stage. In this cohort in the EG, 44% of patients were abstinent at six months compared to 18%in the CG. All patients tended to smoke less after a hospitalisation. Conclusion: The intervention by a smoking cessation nurse during hospitalisation seems effective and is most rewarding in the smokers > 40 years old, and who were well motivated to stop.

After the disappointing experience with lung cancer screening studies based on chest x-ray and/or sputum cytology several decades ago, great expectations are now in place on screening with modern low-dose spiral computed tomography (LDSCT) scan. The prevalence data of one of these randomized studies is reported in this issue of our journal.1Pedersen JH Ashraf H Dirksen A et al.The Danish randomized lung cancer CT screening trial—overall design and results of the prevalence round.J Thorac Oncol. 2009; 4: 608-614Abstract Full Text Full Text PDF PubMed Scopus (327) Google Scholar The definition of screening—adapted from Stedman's Concise dictionary—consists of an examination of a group of usually asymptomatic individuals to detect those with a high probability of having a given disease, typically by means of an inexpensive, safe, and well-performing diagnostic test.2Dirckx JH Stedman's Concise Medical dictionary for the Health Professions. 4th ed. Lippincott Williams & Wilkins, Baltimore2001Google Scholar To reach the real goal—reduction of lung cancer related mortality—four conditions need to be in place: a sensitive test for detection of small lesions; small lesions need to be associated with truly early stage disease; an effective treatment is available for these lesions; and acceptable morbidity and cost of the screening tool. Screening for lung cancer by LDSCT has been studied extensively in the past decade, and it looks like this might become the long awaited tool. Looks like, because until today, all the available data have been gathered from many—some even very large—cohort studies, but not yet from large randomized controlled trials. These nonrandomized studies show promising results. In the I-ELCAP trial in asymptomatic smokers or past smokers, for instance, 85% of lung cancers were detected in stage I, and the estimated 10-year survival rate of patients whose stage I lung cancer was removed by surgery was 92%.3Henschke CI Yankelevitz DF Libby DM et al.Survival of patients with stage I lung cancer detected on CT screening.N Engl J Med. 2006; 355: 1763-1771Crossref PubMed Scopus (1465) Google Scholar This suggests that LDSCT is a sensitive tool that is able to detect small lesions associated with truly early stage, for which an effective therapy is available. This does not prove, however, that the strategy results in a reduction of lung cancer related mortality. This evidence can only come from the eagerly awaited results from two large randomized controlled trials that are currently running, the American National Lung Screening Trial, started in 2002, and the Dutch-Belgian Nederlands-Leuvens Longkanker Screening Onderzoek (NELSON) trial, started in 2003.4Clark KW Gierada DS Marquez G et al.Collecting 48,000 CT exams for the Lung Screening study of the National Lung Screening trial.J Diag Imaging. September 2008; https://doi.org/10.1007/s10278-008-9145-9Crossref Scopus (18) Google Scholar, 5Van Iersel CA De Koning HJ Draisma G et al.Risk-based selection from the general population in a screening trial: selection criteria, recruitment and power for the Dutch-Belgian randomised lung cancer Multi-slice CT screening trial (NELSON).Int J Cancer. 2007; 120: 868-874Crossref PubMed Scopus (425) Google Scholar The long-term postscreening follow-up on the primary end point of reduction of lung cancer mortality in participants in the National Lung Screening Trial and the NELSON trial is planned to end in 2009 and 2014, respectively. Until then, the "pro and con debate" on CT screening for lung cancer will probably remain a "battle over lung scans."6Marshall E A bruishing battle over lung scans.Science. 2008; 320: 600-603Crossref PubMed Scopus (18) Google Scholar, 7Twombly R Lung cancer screening debate continues despite international CT study results.J Natl Cancer Inst. 2007; 99: 190-195Crossref PubMed Scopus (12) Google Scholar Some other, smaller RCTs have been setup in Italy, France, and in Denmark, the Danish randomized lung cancer CT screening trial (DLCST), whose first round or prevalence CT results are reported now.1Pedersen JH Ashraf H Dirksen A et al.The Danish randomized lung cancer CT screening trial—overall design and results of the prevalence round.J Thorac Oncol. 2009; 4: 608-614Abstract Full Text Full Text PDF PubMed Scopus (327) Google Scholar This DLCST has been designed in accordance with the NELSON trial to allow pooling of both study data together once both screening trials will have been finished. The combined data on lung cancer-specific mortality of around 20,000 included individuals will then give these two CT lung cancer screening trials an 80% power to show a lung cancer mortality reduction of at least 25%.5Van Iersel CA De Koning HJ Draisma G et al.Risk-based selection from the general population in a screening trial: selection criteria, recruitment and power for the Dutch-Belgian randomised lung cancer Multi-slice CT screening trial (NELSON).Int J Cancer. 2007; 120: 868-874Crossref PubMed Scopus (425) Google Scholar Although waiting for the final results of these trials' follow-up period by the year 2014, it is interesting to compare the first round CT screening results of the DLCST to the previously reported results from other series. The lung cancer detection rate of 0.83% (17 cases of lung cancer in 2052 participants) is lower than previously reported in nonrandomized CT screening trials. As pointed out by the authors, this result is not easily explainable at this time, and results of the lung cancer prevalence in their control group also need to be awaited. What is probably more important is the stage distribution of lung cancers that were detected in the randomized Danish trial: only 53% of screen detected lung cancers were stage I, definitely lower than in the International Early Lung Cancer Action Program trial. Around 65% of all screen-detected lung cancers in the DLCST could be surgically resected. How this all will translate into better prognosis, improved lung cancer survival and, most importantly, lower lung cancer mortality figures for the screening participants is very difficult to predict at present. Even if lung cancer screening with LDSCT would become a validated screening method, its applicability in large parts of the world will be debated, for its safety and cost-effectiveness. One issue will be the most appropriate algorithm to be used for further diagnosis of noncalcified nodules that will often be detected on LDSCT, to reduce the number of invasive tests for benign nodules, and whether this algorithm can be improved by, e.g., use of more precise estimations of growth by computer-aided volumetric measurements, or by selective use of 18Fluorodeoxyglucose-positron emission tomography scan.8Pastorino U Bellomi M Landoni C et al.Early lung-cancer detection with spiral CT and positron emission tomography in heavy smokers: 2-year results.Lancet. 2003; 362: 593-597Abstract Full Text Full Text PDF PubMed Scopus (402) Google Scholar Another question will be the exact definition of a high-risk population to target for lung cancer screening. Lung cancer risk models (such as the Liverpool Lung Project model) combining different epidemiological risk factors besides smoking history have been developed and will also need to be further validated.9Cassidy A Myles J van Tongeren M et al.The LLP risk model: an individual risk prediction model for lung cancer.Br J Cancer. 2008; 98: 270-276Crossref PubMed Scopus (328) Google Scholar Until that time, we strongly endorse different international medical society guidelines that screening for lung cancer in asymptomatic individuals with LDSCT is not ready for widespread clinical practice in 2009.10Smith RA Cokkinides V Brawley OW Cancer screening in the United States, 2009: A review of current American Cancer Society guidelines and issues in cancer screening.CA Cancer J Clin. 2009; 59: 27-41Crossref PubMed Scopus (370) Google Scholar, 11Bach P Silvestri GA Hanger M Jett JR Screening for lung cancer. ACCP evidence-based clinical practice guidelines (2nd edition).Chest. 2007; 132: 69S-77SCrossref PubMed Scopus (147) Google Scholar Recently reported efforts of total body screening of healthy individuals with CT scan, magnetic resonance imaging, or even FDG-positron emission tomography scan12Nishizawa S Kojima S Teramukai S et al.Prospective evaluation of whole-body cancer screening with multiple modalities including [18F]Fluorodeoxyglucose positron emission tomography in a healthy population: a preliminary report.J Clin Oncol. March 2009; https://doi.org/10.1200/JCO.2008.18.223 8Crossref PubMed Google Scholar similarly belong in clinical trial settings at present. It would be an error to promote LDSCT screening for lung cancer as an evidence-based screening method at present, but we all have hopes that—in a few years' time—we do not need to conclude that this whole idea of trials was an error. Lung cancer continues to be the leading cause of cancer deaths, and a successful strategy to reduce lung cancer mortality is desperately needed. Although awaiting the final results of the randomized trials, it would certainly be an error not to address the main cause of lung cancer, tobacco smoking. What we know for sure at present is that ignoring or not sufficiently addressing the need for smoking cessation of patients in our lung cancer screening trials and in our daily clinical practice, really would be the unforgivable error. We should avoid it, by using counseling and smoking cessation aids whenever possible.

Background: The TNM staging reflects the anatomic extent of lung cancer and estimates the survival expectation. Addition of FDG-PET to conventional staging (CS) improves accuracy, but few data have described the impact of this on long-term survival in relation to treatment. Objectives: To study the influence of FDG-PET on long-term outcome. Methods: Long-term outcome data of patients were retrieved out of previously published PET studies of the Leuven Lung Cancer Group. All patients had a potential for radical treatment, and at least 5-year follow-up data. Patients were dichotomized in early (I–IIIA) versus late (IIIB–IV) stages. Results: A first analysis – comparison of the 2 staging algorithms, CS alone versus CS+PET – confirmed the better staging capabilities of the latter. A second analysis, focusing on discordant findings and interaction of both staging algorithms, demonstrated that patients with early stage on PET did well, while those with late stage on PET did poorly, irrespective of findings on CS. The third analysis focused on the relation between treatment choices at the multidisciplinary board and outcome, which is especially relevant in patients with discordant finding on CS and CS+PET. From all radically treated patients, only those with early stage on CS+PET had a good outcome, but not those with early stage on CS and an unexplained late stage finding on PET. Conclusion: This long-term follow-up analysis confirms that addition of PET to CS results in better stage designation and prognosis. Additionally, discordant findings between CS and CS+PET should be considered relevant, with need for cytological/histological examination.

1508 Background: Early lung cancer detection with X-ray was unable to reduce lung cancer mortality. Low-dose spiral CT screening cohort studies have demonstrated that lung cancer can be detected at an early stage in a high percentage. Unknown is whether spiral CT screening will lead to a lung cancer mortality reduction and if lung cancer screening will be cost-effective. These questions will be addressed in the randomized Dutch-Belgian-Danish lung cancer screening trial (NELSON) of 20,000 participants. Methods: To identify high-risk subjects a health questionnaire was sent to 600,000 men and women aged 50–74 in seven Dutch public health districts and 14 municipalities in Belgium. Current and former smokers (quit < 10 years) who smoked at least 16 cigarettes a day for at least 26 years or at least 11 cigarettes a day for at least 31 years were invited. After receiving informed consent, participants were randomized between screen-arm and control-arm (usual care). CT scans are performed with 16-detector mult...

PURPOSE:Surgical resection in patients with stage IIIA-N2 non-small-cell lung cancer (NSCLC) is usually reserved for patients with mediastinal downstaging after induction chemotherapy (IC). However, clinical restaging is often inaccurate, and there are insufficient data to conclude that all patients with persistent mediastinal disease will not benefit from surgery, or that all patients with mediastinal clearance benefit from surgery. We created a data-based restaging strategy combining morphometric tissue analysis of mediastinal lymph nodes (LNs) and 18-fluoro-2-deoxy-glucose positron emission tomography (FDG-PET) response monitoring in the primary tumor.PATIENTS AND METHODS:Baseline and repeat FDG-PET after IC, as well as complete resection specimens of both mediastinal LNs and primary tumor, were available in 30 patients. Histologic response grading was performed by means of conventional morphometric procedures. Mediastinal response grading combined with the percentage decrease of maximum standardized uptake value (SUV(max)) on the primary tumor was correlated with survival.RESULTS:Patients with persistent major mediastinal LN involvement have a 5-year overall survival rate of 0%. The 5-year overall survival rate for patients with cleared or persistent minor mediastinal LN involvement was significantly higher in patients with a more than 60% decrease in SUV(max) on the primary tumor as compared with patients with a less than 60% decrease in SUV(max) (62% v 13%; log-rank P = .002).CONCLUSION:These data may suggest that (1) persistent mediastinal disease after IC does not always exclude favorable outcome after surgery; (2) serial FDG-PET may select surgical candidates among patients with mediastinal downstaging or persistent minor disease; (3) persistent major mediastinal disease has a poor prognosis and such patients should not be considered for surgery.

We report the case of an unexpected 18F-fluorodeoxyglucose-avid lesion in the right lower abdomen in a patient with otherwise "very limited" (T1N0) small cell lung cancer (SCLC). Additional imaging and endoscopic studies showed no abnormality. The patient was treated for presumed very limited disease SCLC, with resection, adjuvant chemotherapy, and prophylactic brain irradiation. Follow-up fusion positron emission tomography-computed tomography revealed an unusual SCLC complication.

Surgical resection in patients with stage IIIA-pN2 NSCLC is usually reserved for patients with downstaging of mediastinal lymph nodes (LNs) after induction therapy. However, clinical restaging after induction therapy is often inaccurate. Moreover, there are insufficient data to exclude all patients with persistent mediastinal disease from surgery. Assuming baseline endoscopic mediastinal staging in the near future, we explored a restaging strategy combining tissue analysis of mediastinal LNs obtained post-induction, and FDG-PET for monitoring response in the primary tumour. Baseline and repeat PET after three cycles of induction chemotherapy, as well as complete resection specimens of both mediastinal LNs and primary tumour, were available in 30 patients out of our prospective database of surgical multimodality therapy in IIIA-pN2 NSCLC patients (period: 04-1995 to 06-2002; ref. Lorent et al, Ann Oncol 15:1645, 2004). In these 30 patients, histological tumour grading of both LNs and primary tumour was performed by means of conventional morphometric procedures based on a point counting technique, as described by Gundersen (ref. Gundersen et al, APMIS 96:379, 1988). Three morphometric grading groups were considered: mediastinal downstaging (clearance of mediastinal disease), ‘minor' residual mediastinal disease (<10% viable tumour cells) and ‘major' residual mediastinal disease (>10% viable tumour cells). These mediastinal morphometric grading groups, together with the percent decrease of SUVmax of the primary tumour, were correlated with survival to establish an algorithm for prognostic restaging. The median and 5-year overall survival for the 30 patients was 36.4 months and 30%, respectively. The median decrease of SUVmax on the primary tumour between baseline and repeat PET scan was 60%, and this value was used as cut-off. A >60% decrease compared to a <60% decrease of SUVmax resulted in a median survival time of 57.5 versus 18.3 months and 5-year overall survival 47% versus 13%, respectively; log-rank p=0.01, HR 0.34 (95%CI 0.12-0.74). No patient with ‘major' residual mediastinal disease survived 5 years, while 5-year overall survival was 43% both in patients with mediastinal clearance as well as in those with ‘minor' residual mediastinal disease. Combined prognostic stratification resulted in two subgroups: patients with downstaging or persistent ‘minor' mediastinal LN involvement and >60% decrease of SUVmax on the primary tumour were allocated as ‘Good Prognosis group' (n=13). Patients with downstaging or persistent ‘minor' mediastinal LN involvement and <60% decrease of SUVmax on the primary tumour were pooled with patients having persistent ‘major' mediastinal LN involvement and allocated as ‘Poor Prognosis group' (n=17). The median and 5-year overall survival for these Good and Poor Prognosis groups were “not reached” versus 18.3 months, and 62% versus 6%, respectively; log-rank p<0.0001, HR 0.18 (95%CI 0.06-0.38). These data suggest: (1) persistent ‘minor' pN2-disease following induction chemotherapy does not always exclude favourable outcome after surgery; (2) serial FDG-PET is able to select surgical candidates amongst patients with persistent ‘minor' pN2-disease or with mediastinal clearance; (3) persistent ‘major' pN2-disease has a poor prognosis and should not be considered for surgery.

The interpretation of gap fraction in terms of LAI from optical in situ methods is based on light extinction models, which link LAI and canopy architecture to the penetration of solar radiation through the canopy. The selection of the most appropriate LAI inversion model to calculate LAI from gap fraction data extracted from high-dynamic range digital hemispherical photographs is a crucial step in the development of a standardized protocol for the digital hemispherical camera. Three currently available light extinction models, all based on different assumptions concerning the foliage distribution, were tested in this study with respect to the appropriateness of the respective canopy elements distribution theories for LAI inversion. Reference LAI values (L_exact) from virtual 3-D forest stands were compared to indirectly measured LAI values (L_HP) obtained from simulated hemispherical photographs taken in the virtual stands. A regression between the reference LAI and the indirectly measured LAI selected the Markov model as most optimal inversion model. Fractal dimension (FD) was tested as a correction parameter for the deviation of forest canopies from assumed theoretical foliage distributions. Inclusion of a FD correction parameter improved the goodness-of-fit of the indirect LAI versus reference LAI from 86.7% to 97%. Virtual 3-D forest stands generated by L-systems provided a very appropriate validation environment for this research. Finally, the validity of the FD correction parameter was also tested with actual forest stand data.

Sustainable forest management requires accurate and up‐to‐date information, which can nowadays be obtained using digital earth observation technology. This paper introduces a modified change vector analysis (mCVA) approach and conceptually contrasts it against traditional CVA. The results of a comparative study between this change detection algorithm and three other widely used change detection algorithms: standardized differencing, ratioing and selective principal component analysis are summarized. Landsat Thematic Mappper (TM) imagery and detailed change maps of a forested area in Northern Minnesota were used. Change indicators (vegetation indices) were grouped into three conceptually independent categories corresponding to soil, vegetation and moisture characteristics. Change periods of two, four and six years were considered. All change detection outputs were multidimensional and of a continuous nature, and could therefore be subjected to a supervised maximum likelihood algorithm using identical data training sets. Change extraction accuracies were determined by computing overall accuracy and Kappa coefficients of agreement against independent reference datasets. The mCVA outperformed the three other change detection methods in all cases, and we have shown that there is a clear advantage in running mCVA with three change indicator inputs where each input comes from a different change indicator category. Further validations with more detailed reference data are needed to improve this method and test its performance for other types of change events.

Thresholding is a central part of the analysis of hemispherical images in terms of gap fraction and leaf area index (LAI), and the selection of optimal thresholds has remained a challenge over decades. The need for an objective, automatic, operator-independent thresholding method has long been of interest to scientists using hemispherical photography. This manuscript deals with the comparison of a wide variety of different well-known automatic thresholding techniques against the subjective manual method, using high-dynamic range digital hemispherical photographs. The performance of the different thresholding methods was evaluated based on: (1) visual inspection by means of a multi-criteria decision system and (2) quantitative analysis of the methods’ sensitivity to an overall performance criterium. The automatic Ridler clustering method proved to be the most robust thresholding method for various canopy structure conditions. This automatic method might be the best solution for a fast, reliable and objective use of hemispherical photographs for gap fraction and LAI estimation in forest stands, given that the threshold setting is no longer manually performed. The fine-tuning potential of local thresholding methods to better address particular photographic limitations (e.g. over-exposure in a certain image region) is also presented.

Traditionally, the validation of a classified multispectral image only quantifies its correspondence to ground reference data containing thematic information generalized at the stand level, with stands represented as vector polygons. Little is known of the accuracy of such classifications at a scale below the stand. This study presents a methodology to assess classification accuracy at pixel level, i.e. sub-polygon, where the classification procedure is embedded in a change detection environment. A new type of reference data (Metatruth Image) was generated based on the integration of the outputs of various independent change detection procedures. The integration consisted of calculating for each pixel a probability distribution or pixel purity index for each change class by independent change detection procedures, defined by the number of times the pixel has been classified as a certain change class. First, the relationship between purity and accuracy was successfully validated. Next, the Metatruth Image was created based on 'high purity pixels'. Performing traditional accuracy assessment on the outputs of individual change detection procedures using the Metatruth Image as reference dataset, demonstrated that former outputs identified change events accurately at pixel level. As a consequence, traditional accuracy assessment at polygon level underestimates the true accuracy at pixel level of the change detection procedure in a systematic way with differences in kappa coefficients of agreement around 20%.

Rapid, reliable and objective estimations of leaf area index (LAI) are essential for numerous studies of atmosphere-vegetation interaction, as LAI is very often a critical parameter in process-based models of vegetation canopy response to global environmental change. This paper reviews current knowledge concerning the use of direct and indirect methods for LAI determination. The value of optical LAI measurements by means of hemispherical photography has already been demonstrated in previous studies. As clumping seems to be the main factor causing errors in indirect LAI estimation, we suggest that the use of a digital camera with high dynamic range has the potential to overcome a number of described technical problems related to indirect LAI estimation. Further testing and defining of a standardised field protocol for digital hemispherical photography is however needed to improve this technique to achieve the standards of an ideal device. (C) 2003 Elsevier B.V. All rights reserved.